1 2017-11-17 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/83017
4 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5 (gen_parallel_loop): Properly count iterations.
6 (parallelize_loops): Handle loop->can_be_parallel independent
7 of flag_loop_parallelize_all. Make static profitability test match
9 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
10 * invoke.texi (parloops-min-per-thread): Document.
12 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
14 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
15 upstreaming review comments.
17 2017-11-17 Tamar Christina <tamar.christina@arm.com>
19 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
20 with fast unaligned access.
21 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
23 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
26 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
28 2017-11-17 Richard Biener <rguenther@suse.de>
30 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
31 folding of references.
33 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
36 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
37 of replacing call to strncmp with corresponding call to strcmp when
40 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
42 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
43 option prefer-avx256 for skylake-avx512 configuration.
44 * config/i386/i386.c (ix86_option_override_internal): Ditto.
45 (get_builtin_code_for_version): Ditto.
47 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
48 Monk Chiang <sh.chiang04@gmail.com>
50 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
51 (FIXED_REGISTERS): Reserve more register numbers.
52 (CALL_USED_REGISTERS): Likewise.
53 (REG_ALLOC_ORDER): Likewise.
54 (REG_CLASS_CONTENTS): Likewise.
55 (REGISTER_NAMES): Likewise.
57 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
58 Kito Cheng <kito.cheng@gmail.com>
60 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
62 * config/nds32/iterators.md: Add vector mode iterators and attributes.
64 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
66 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
67 parameter list for vec_splats.
69 2017-11-16 Joseph Myers <joseph@codesourcery.com>
71 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
73 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
76 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
79 * doc/install.texi: Document incompatibility of Perl >=5.6.26
80 with the required version of automake 1.11.6.
82 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
84 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
86 (DU_C2_3_power9): Correct reservation combinations.
87 (FP_DIV_power9, VEC_DIV_power9): New.
88 (power9-alu): Split out rotate/shift...
89 (power9-rot): ...to here, correct dispatch resource.
90 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
92 (power9-fp): Correct latency.
93 (power9-sdiv): Add div/sqrt resource.
94 (power9-ddiv): Correct latency, add div/sqrt resource.
95 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
96 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
98 (power9-qpdiv, power9-qpmul): Adjust resource usage.
100 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
102 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
103 switch statement mapping KF built-ins to TF built-ins if we don't
104 have the proper ISA 3.0 assembler support.
106 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
108 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
109 (lower_emutls_function_body): Do not compute it.
111 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
113 * ipa-split.c (split_bb_info): Turn time to sreal.
114 (split_point): Likewise.
115 (dump_split_point): Likewise.
116 (fine_split_points): Likewise.
117 (execute_split_functions): Only zero split_bbs; turn time to sreals.
119 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
121 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
124 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
126 * predict.c (combine_predictions_for_bb): Preserve zero predicted
128 (expensive_function_p): Remove useless assert.
129 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
131 2017-11-16 Martin Sebor <msebor@redhat.com>
133 PR tree-optimization/82588
134 PR tree-optimization/82583
135 * tree-vrp.c (check_array_ref): Handle flexible array members,
136 string literals, and inner indices.
137 (search_for_addr_array): Add detail to diagnostics.
139 2017-11-16 Nathan Sidwell <nathan@acm.org>
143 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
144 (SET_DECL_ASSEMBLER_NAME): Forward to
145 overwrite_decl_assembler_name.
146 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
147 (overwrite_decl_assembler_name): Declare.
148 * tree.c (overwrite_decl_assembler_name): New.
149 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
150 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
151 (LANG_HOOKS_INITIALIZER): Add it.
152 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
153 * langhooks.c (lhd_set_decl_assembler_name): Use
154 SET_DECL_ASSEMBLER_NAME.
155 (lhd_overwrite_decl_assembler_name): Default implementation.
157 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
158 Jackson Woodruff <jackson.woodruff@arm.com>
160 PR tree-optimization/71026
161 * match.pd: Canonicalize constant multiplies in division.
163 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
165 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
166 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
167 Remove OPT_fomit_frame_pointer entry.
168 * common/config/arc/arc-common.c: Likewise.
169 * common/config/arm/arm-common.c: Likewise.
170 * common/config/avr/avr-common.c: Likewise.
171 * common/config/c6x/c6x-common.c: Likewise.
172 * common/config/cr16/cr16-common.c: Likewise.
173 * common/config/cris/cris-common.c: Likewise.
174 * common/config/epiphany/epiphany-common.c: Likewise.
175 * common/config/fr30/fr30-common.c: Likewise.
176 * common/config/frv/frv-common.c: Likewise.
177 * common/config/ia64/ia64-common.c: Likewise.
178 * common/config/iq2000/iq2000-common.c: Likewise.
179 * common/config/lm32/lm32-common.c: Likewise.
180 * common/config/m32r/m32r-common.c: Likewise.
181 * common/config/mcore/mcore-common.c: Likewise.
182 * common/config/microblaze/microblaze-common.c: Likewise.
183 * common/config/mips/mips-common.c: Likewise.
184 * common/config/mmix/mmix-common.c: Likewise.
185 * common/config/mn10300/mn10300-common.c: Likewise.
186 * common/config/nios2/nios2-common.c: Likewise.
187 * common/config/pa/pa-common.c: Likewise.
188 * common/config/pdp11/pdp11-common.c: Likewise.
189 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
190 * common/config/riscv/riscv-common.c: Likewise.
191 * common/config/rs6000/rs6000-common.c: Likewise.
192 * common/config/rx/rx-common.c: Likewise.
193 * common/config/s390/s390-common.c: Likewise.
194 * common/config/sh/sh-common.c: Likewise.
195 * common/config/sparc/sparc-common.c: Likewise.
196 * common/config/tilegx/tilegx-common.c: Likewise.
197 * common/config/tilepro/tilepro-common.c: Likewise.
198 * common/config/v850/v850-common.c: Likewise.
199 * common/config/visium/visium-common.c: Likewise.
200 * common/config/xstormy16/xstormy16-common.c: Likewise.
201 * common/config/xtensa/xtensa-common.c: Likewise.
202 * invoke.texi (-fomit-frame-pointer): Update documentation.
204 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
206 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
208 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
210 * cfg.c (scale_bbs_frequencies_int,
211 cale_bbs_frequencies_gcov_type): Remove.
212 * cfg.h (scale_bbs_frequencies_int,
213 cale_bbs_frequencies_gcov_type): Remove.
215 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
217 * tree-ssa-loop-manip.c
218 (scale_dominated_blocks_in_loop): Update to profile counts.
219 (tree_transform_and_unroll_loop): Likewise.
221 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
223 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
224 scale_bbs_frequencies_int.
226 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
228 * final.c (compute_alignments): Use counts rather than frequencies.
230 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
232 * cfgloopanal.c: Include sreal.h
233 (average_num_loop_insns): Use counts and sreal for accounting.
235 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
237 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
240 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
241 Kito Cheng <kito.cheng@gmail.com>
243 * config/nds32/constraints.md: Provide more constraints.
244 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
245 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
246 support constraints usage.
248 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
250 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
252 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
253 Kito Cheng <kito.cheng@gmail.com>
255 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
256 * config/nds32/nds32.opt: Refine the layout.
257 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
258 TARGET_EXT_STRING): Support new options.
259 * config/nds32/nds32.h: Likewise.
260 * config/nds32/nds32.md: Likewise.
261 * config/nds32/nds32-predicates.c: Likewise.
262 * config/nds32/constraints.md: Likewise.
263 * common/config/nds32/nds32-common.c: Likewise.
265 2017-11-16 Julia Koval <julia.koval@intel.com>
268 * config/i386/gfniintrin.h: Add sse check.
269 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
271 2017-11-16 Julia Koval <julia.koval@intel.com>
273 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
274 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
275 (ix86_handle_option): Handle -mavx512vbmi2.
276 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
277 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
278 * config/i386/i386-c.c (__AVX512VBMI2__): New.
279 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
280 (ix86_valid_target_attribute_inner_p): Ditto.
281 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
282 * config/i386/i386.opt (mavx512vbmi2): New option.
283 * doc/invoke.texi: Add new option.
285 [2017-11-16 Julia Koval <julia.koval@intel.com>
287 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
288 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
289 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
290 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
291 _mm512_gf2p8mul_epi8): New intrinsics.
292 * config/i386/i386-builtin-types.def
293 (V64QI_FTYPE_V64QI_V64QI): New type.
294 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
295 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
296 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
297 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
298 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
299 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
301 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
303 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
304 explicitly as a stream of bytes.
306 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
308 * config/rs6000/altivec.h (vec_xst_be): New #define.
309 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
310 and externalize from *altivec_vperm_<mode>_internal.
311 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
313 (XL_BE_V8HI): Likewise.
314 (XL_BE_V4SI): Likewise.
315 (XL_BE_V4SI): Likewise.
316 (XL_BE_V2DI): Likewise.
317 (XL_BE_V4SF): Likewise.
318 (XL_BE_V2DF): Likewise.
319 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
320 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
321 all array entries with these keys: VSX_BUILTIN_VEC_XL,
322 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
323 VSX_BUILTIN_VEC_XST_BE.
324 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
325 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
327 (altivec_init_builtins): Replace conditional calls to def_builtin
328 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
329 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
330 with unconditional calls. Remove calls to def_builtin for
331 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
332 __builtin_vec_xst_be.
333 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
334 to define_expand, and add alternate RTL generation for P8.
335 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
337 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
338 add alternate RTL generation for P8.
339 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
340 vsx_ld_elemrev_v16qi.
341 (vsx_st_elemrev_v8hi): Convert define_insn
342 to define_expand, and add alternate RTL generation for P8.
343 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
345 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
346 add alternate RTL generation for P8.
347 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
348 vsx_st_elemrev_v16qi.
350 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
353 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
354 TARGET_AVX512ER check.
355 (ix86_option_override_internal): Set MASK_VZEROUPPER if
356 neither -mzeroupper nor -mno-zeroupper is used and
357 TARGET_EMIT_VZEROUPPER is set.
358 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
359 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
361 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
363 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
364 folding of vector compares.
365 (fold_build_vec_cmp): New helper function.
366 (fold_compare_helper): New helper function.
367 (builtin_function_type): Add compare builtins to the list of functions
368 having unsigned arguments. Cosmetic updates to comment indentation.
369 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
370 the not+eq combination.
372 2017-11-15 Bin Cheng <bin.cheng@arm.com>
374 PR tree-optimization/82726
375 PR tree-optimization/70754
376 * tree-predcom.c (order_drefs_by_pos): New function.
377 (combine_chains): Move code setting has_max_use_after to...
378 (try_combine_chains): ...here. New parameter. Sort combined chains
379 according to position information.
380 (tree_predictive_commoning_loop): Update call to above function.
381 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
383 2017-11-15 Bin Cheng <bin.cheng@arm.com>
385 PR tree-optimization/82726
387 2017-01-23 Bin Cheng <bin.cheng@arm.com>
389 PR tree-optimization/70754
390 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
391 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
392 combined stmt before it if not NULL.
393 (combine_chains): Process refs reversely and compute dominance point
397 2017-02-23 Bin Cheng <bin.cheng@arm.com>
399 PR tree-optimization/79663
400 * tree-predcom.c (combine_chains): Process refs in reverse order
401 only for ZERO length chains, and add explaining comment.
403 2017-11-15 Tamar Christina <tamar.christina@arm.com>
405 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
406 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
407 * doc/invoke.texi (ARM Options): Add armv8.3-a.
409 2017-11-15 Tamar Christina <tamar.christina@arm.com>
411 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
413 2017-11-15 Martin Liska <mliska@suse.cz>
415 * tree-cfg.c (pass_warn_function_return::execute):
416 Compare warn_return_type for greater than zero.
418 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
422 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
423 to return true on Xeon and not on Xeon Phi.
424 (ix86_check_avx256_register): Changed to ...
425 (ix86_check_avx_upper_register): ... this. Add extra check for
426 VALID_AVX512F_REG_OR_XI_MODE.
427 (ix86_avx_u128_mode_needed): Changed
428 ix86_check_avx256_register to ix86_check_avx_upper_register.
429 (ix86_check_avx256_stores): Changed to ...
430 (ix86_check_avx_upper_stores): ... this. Changed
431 ix86_check_avx256_register to ix86_check_avx_upper_register.
432 (ix86_avx_u128_mode_after): Changed
433 avx_reg256_found to avx_upper_reg_found. Changed
434 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
435 (ix86_avx_u128_mode_entry): Changed
436 ix86_check_avx256_register to ix86_check_avx_upper_register.
437 (ix86_avx_u128_mode_exit): Ditto.
438 * config/i386/i386.h: (host_detect_local_cpu): New define.
440 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
442 * config/arm/xgene1.md (xgene1): Split into automatons
443 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
444 (xgene1_f_load): Adjust reservations and/or types.
445 (xgene1_f_store): Likewise.
446 (xgene1_load_pair): Likewise.
447 (xgene1_store_pair): Likewise.
448 (xgene1_fp_load1): Likewise.
449 (xgene1_load1): Likewise.
450 (xgene1_store1): Likewise.
451 (xgene1_move): Likewise.
452 (xgene1_alu): Likewise.
453 (xgene1_simd): Likewise.
454 (xgene1_bfm): Likewise.
455 (xgene1_neon_load1): Likewise.
456 (xgene1_neon_store1): Likewise.
457 (xgene1_neon_logic): Likewise.
458 (xgene1_neon_st1): Likewise.
459 (xgene1_neon_ld1r): Likewise.
460 (xgene1_alu_cond): Added.
461 (xgene1_shift_reg): Likwise.
462 (xgene1_bfx): Likewise.
463 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
465 2017-11-15 Jakub Jelinek <jakub@redhat.com>
468 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
470 (can_widen_mult_without_libcall): New function.
471 (expand_mul_overflow): If only checking unsigned mul overflow,
472 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
473 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
474 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
475 (expand_DIVMOD): Formatting fix.
476 * expmed.h (expand_mult): Add NO_LIBCALL argument.
477 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
478 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
480 PR tree-optimization/82977
481 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
482 constructed temporary to strlen_to_stridx.put.
484 2017-11-15 Martin Liska <mliska@suse.cz>
486 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
487 * configure: Regenerate.
489 2017-11-15 Martin Liska <mliska@suse.cz>
492 * config/sh/sh-mem.cc: Use proper probability for
495 2017-11-14 Jeff Law <law@redhat.com>
497 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
498 the red zone for stack_clash_protection_final_dynamic_probe targets
499 when the total dynamic stack size is zero bytes.
501 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
502 blocks is post order.
504 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
506 * dumpfile.h (TDF_COMPARE_DEBUG): New.
507 * final.c (rest_of_clean_state): Set it for the
508 -fcompare-debug dump.
509 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
510 class when TDF_COMPARE_DEBUG is set.
512 * dwarf2out.c (gen_producer_string): Discard
515 2017-11-15 Joseph Myers <joseph@codesourcery.com>
518 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
519 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
520 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
521 (__TGMATH_REAL_2_3): Remove macros.
522 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
523 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
525 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
527 (remquo): Define using __TGMATH_REAL_3.
529 2017-11-14 Jeff Law <law@redhat.com>
531 * vr-values.c: New file with contents extracted from tree-vrp.c.
532 * Makefile.in (OBJS): Add vr-values.o
533 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
534 (set_value_range, set_and_canonicalize_value_range): Likewise.
535 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
536 (value_range_constant_singleton, symbolic_range_p): Likewise.
537 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
538 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
539 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
540 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
541 (find_case_label_range, find_case_label_index): Likewise.
542 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
543 (range_int_cst_singleton_p, value_inside_range): Likewise.
544 (get_single_symbol): Likewise.
545 (switch_update): Move structure definition here.
546 (to_remove_edges, to_update_switch_stmts): Provide externs.
547 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
548 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
549 (vrp_val_is_min, set_value_range): Likewise.
550 (set_and_canonicalize_value_range, copy_value_range): Likewise.
551 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
552 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
553 (range_is_nonnull, range_int_cst_p): Likewwise.
554 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
555 (get_single_symbol, operand_less_p): Likewise
556 (compare_values_warnv, compare_values): Likewise.
557 (value_inside_range, value_range_constant_singleton): Likewise.
558 (zero_nonzero_bitgs_from_vr): Likewise.
559 (extract_range_from_binary_expr_1): Likewise.
560 (overflow_comparison_p): Likewise.
561 (to_remove_edges, to_update_switch_stmts): Likewise.
562 (find_case_label-index, find_case_label_range): Likewise.
563 (switch_update, set_value_range_to_nonnegative): Remove.
564 (set_value_range_to_truthvalue): Likewise.
565 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
566 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
567 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
568 (find_case_label_ranges, test_for_singularity): Likewise.
569 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
570 (x_vr_values): Move to its remaining use site.
572 2017-11-10 Jeff Law <law@redhat.com>
574 * vr-values.h (VR_INITIALIZER): Move #define here.
575 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
576 * Makefile.in (OBJS): Add tree-evrp.o
577 * tree-vrp.h (assert_info): Move structure definition here.
578 (set_value_range_to_varying): Prototype.
579 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
580 (infer_value_range, register_edge_assert_for): Likewise.
581 (stmt_interesting_for_vrp): Likewise.
582 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
583 (set_value_range_to_varying): No longer static.
584 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
585 (infer_value_range, register_edge_assert_for): Likewise.
587 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
589 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
590 generate the XXBRD instruction.
592 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
593 (rs6000_builtin_type_compatible): Treat _Float128 and long double
594 as being compatible if -mabi=ieeelongdouble.
595 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
596 to setup float128 built-ins with hardware support.
597 (BU_FLOAT128_HW_2): Likewise.
598 (BU_FLOAT128_HW_3): Likewise.
599 (BU_FLOAT128_HW_VSX_1): Likewise.
600 (BU_FLOAT128_HW_VSX_2): Likewise.
601 (scalar_extract_expq): Change float128 built-in functions to
602 accommodate having both KFmode and TFmode functions. Use the
603 KFmode variant as the default.
604 (scalar_extract_sigq): Likewise.
605 (scalar_test_neg_qp): Likewise.
606 (scalar_insert_exp_q): Likewise.
607 (scalar_insert_exp_qp): Likewise.
608 (scalar_test_data_class_qp): Likewise.
609 (sqrtf128_round_to_odd): Delete processing the round to odd
610 built-in functions as special built-in functions, and define them
611 as float128 built-ins. Use the KFmode variant as the default.
612 (truncf128_round_to_odd): Likewise.
613 (addf128_round_to_odd): Likewise.
614 (subf128_round_to_odd): Likewise.
615 (mulf128_round_to_odd): Likewise.
616 (divf128_round_to_odd): Likewise.
617 (fmaf128_round_to_odd): Likewise.
618 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
619 support for KFmode and TFmode xststdcqp calls.
620 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
621 point, switch the built-in handlers for the get/set float128
622 exponent, get float128 mantissa, float128 test built-ins, and the
623 float128 round to odd built-in functions. Eliminate creating the
624 float128 round to odd built-in functions as special built-ins.
625 (rs6000_init_builtins): Eliminate special creation of the float128
626 round to odd built-in functions.
627 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
628 function insns to support both TFmode and KFmode variants.
629 (xsxsigqp_<mode>): Likewise.
630 (xsiexpqpf_<mode>): Likewise.
631 (xsiexpqp_<mode>): Likewise.
632 (xststdcqp_<mode>): Likewise.
633 (xststdcnegqp_<mode>): Likewise.
634 (xststdcqp_<mode>): Likewise.
636 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
638 * tree-ssa-threadupdate.c (compute_path_counts): Remove
639 unused path_in_freq_ptr parameter.
640 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
642 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
644 * ipa-inline.c (edge_badness): Dump sreal frequency.
645 (compute_inlined_call_time): Match natural implementaiton ...
646 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
647 forgotten division by CGRAPH_FREQ_BASE.
649 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
651 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
652 Solaris 11. Update comment.
653 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
655 * config/sol2.h (STARTFILE_SPEC): Likewise.
656 * configure: Regenerate.
658 2017-11-14 Carl Love <cel@us.ibm.com>
660 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
661 le_ and be_ prefixes to swap* variables. Remove
662 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
665 2017-11-14 Jason Merrill <jason@redhat.com>
667 Support GTY((cache)) on hash_map.
668 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
669 (ggc_cache_remove): Override it instead of ggc_mx.
670 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
671 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
672 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
673 (simple_cache_map_traits): Override maybe_mx.
674 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
675 (hash_map): Friend gt_cleare_cache.
676 (gt_cleare_cache): New.
677 * tree.h (tree_cache_traits): New hash_map traits class.
678 (tree_cache_map): New typedef.
680 2017-11-14 Richard Biener <rguenther@suse.de>
682 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
683 paramter and handling.
684 (cleanup_control_flow_bb): Likewise.
685 (cleanup_control_flow_pre): New helper performing a DFS walk
686 to call cleanup_control_flow_bb in PRE order.
687 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
688 via cleanup_control_flow_pre.
690 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
692 * config/aarch64/aarch64-simd.md
693 (aarch64_simd_bsl<mode>_internal): Remove DImode.
694 (*aarch64_simd_bsl<mode>_alt): Likewise.
695 (aarch64_simd_bsldi_internal): New.
696 (aarch64_simd_bsldi_alt): Likewise.
698 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
700 * tracer.c (better_p): Do not compare frequencies.
701 * reg-stack.c (better_edge): Likewise.
702 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
705 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
707 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
708 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
709 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
710 * lto-streamer-in.c (input_function): Use update_max_bb_count.
711 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
712 * predict.c (maybe_hot_frequency_p): Inline to ...
713 (maybe_hot_count_p): ... here; rewrite to counts.
714 (counts_to_freqs): Rename to ...
715 (update_max_bb_count): ... this one.
716 (expensive_function_p): Use counts.
717 (estimate_bb_frequencies): Update.
718 (rebuild_frequencies): Update.
719 * predict.h (counts_to_freqs): Rename to ...
720 (update_max_bb_count): ... this one.
721 * profile.c (compute_branch_probabilities): Add debug info
722 * tree-inline.c (expand_call_inline): Update debug info.
723 (optimize_inline_calls): Use update_max_bb_count..
724 (tree_function_versioning): Use update_max_bb_count..
725 * value-prof.c (gimple_value_profile_transformations):
726 Do not use update_max_bb_count.
728 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
730 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
731 always use frequencies.
733 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
735 * bb-reorder.c: Remove frequencies from comments.
736 (better_edge_p): Use profile counts.
737 (find_traces): Dump profile counts.
738 (rotate_loop): Use profile counts.
739 (find_traces_1_round): Likewise.
740 (connect_better_edge_p): Use counts instead of probabilities for
742 (copy_bb_p): Drop early check for non-0 frequency.
743 (sanitize_hot_paths): Update comments.
745 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
747 * ipa-split.c (struct split_point): Add count.
748 (consider_split): Do not compute incoming frequency; compute incoming
749 count and store it to split_point.
750 (split_function): Set count of the call to split part correctly.
752 2017-11-13 Carl Love <cel@us.ibm.com>
754 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
756 2017-11-13 Tom Tromey <tom@tromey.com>
758 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
760 2017-11-13 Carl Love <cel@us.ibm.com>
762 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
763 Add support for builtins:
764 unsigned int vec_first_{,miss}_match_{,or_eos}index,
765 vector {un,}signed {char,int,short},
766 vector {un,}signed {char,int,short}) arguments.
767 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
768 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
769 Add BU_P9V_AV_2 expansions for the builtins.
770 * config/rs6000/altivec.h (vec_first_match_index,
771 vec_first_mismatch_index, vec_first_match_or_eos_index,
772 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
773 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
774 new extern declaration.
775 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
776 * config/rs6000/vsx.md (first_match_index_<mode>,
777 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
778 first_mismatch_or_eos_index_<mode>): Add define expand.
779 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
780 * doc/extend.texi: Update the built-in documenation file for the new
783 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
785 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
786 into the min/max operations for _Float<N> and _Float<N>X types.
788 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
791 * dwarf2out.c (do_eh_frame): New static variable.
792 (dwarf2out_begin_prologue): Set it.
793 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
795 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
797 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
799 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
801 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
804 * debug.h (dwarf2out_do_eh_frame): Declare.
805 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
806 (dwarf2out_do_frame): Use it.
807 (dwarf2out_do_cfi_asm): Likewise.
808 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
809 (dwarf2out_assembly_start): Likewise.
810 (dwarf2out_begin_prologue): Fix comment.
811 * toplev.c (compile_file): Always call dwarf2out_frame_finish
812 if the target needs either debug or unwind DWARF2 info.
813 * lto-opts.c (lto_write_options): Do not save -fexceptions,
814 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
815 -ftrapping-math, -ftrapv and -fwrapv.
817 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
819 * cgraph.c (cgraph_edge::sreal_frequency): New function.
820 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
821 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
822 (estimate_edge_size_and_time): Likewise.
823 (ipa_merge_fn_summary_after_inlining): Likewise.
824 * ipa-inline.c (cgraph_freq_base_rec): Remove.
825 (compute_uninlined_call_time): Use sreal_frequency.
826 (compute_inlined_call_time): Likewise.
827 (ipa_inline): Do not initialize cgraph_freq_base_rec.
828 * profile-count.c: Include sreal.h.
829 (profile_count::to_sreal_scale): New.
830 * profile-count.h: Forward declare sreal.
831 (profile_count::to_sreal_scale): Declare.
833 2017-11-13 Nathan Sidwell <nathan@acm.org>
835 * diagnostic.c (maybe_line_and_column): New.
836 (diagnostic_get_location_text): Use it.
837 (diagnostic_report_current_module): Likewise.
838 (test_diagnostic_get_location_text): Add tests.
840 2017-11-13 Luis Machado <luis.machado@linaro.org>
842 * doc/md.texi (Specifying processor pipeline description): Fix
843 incorrect latency for the div instruction example.
845 2017-11-13 Jakub Jelinek <jakub@redhat.com>
847 PR tree-optimization/78821
848 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
849 that bit_not_p is the same.
850 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
851 (split_group): Count precisely bit_not_p bits in each statement.
852 (invert_op): New function.
853 (imm_store_chain_info::output_merged_store): Use invert_op to
854 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
855 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
857 2017-11-13 Martin Liska <mliska@suse.cz>
859 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
860 (struct source_info): Likewise.
861 (add_branch_counts): Likewise.
862 (add_line_counts): Likewise.
863 (function_summary): Likewise.
864 (output_intermediate_line): Likewise.
865 (generate_results): Likewise.
867 2017-11-13 Martin Liska <mliska@suse.cz>
869 * gcov.c (struct block_info): Remove typedef for block_t.
870 (struct line_info): Likewise.
871 (line_info::has_block): Likewise.
872 (EXIT_BLOCK): Likewise.
875 (get_cycles_count): Likewise.
876 (process_file): Likewise.
877 (read_graph_file): Likewise.
878 (solve_flow_graph): Likewise.
879 (find_exception_blocks): Likewise.
880 (add_line_counts): Likewise.
881 (accumulate_line_info): Likewise.
882 (output_line_details): Likewise.
884 2017-11-13 Martin Liska <mliska@suse.cz>
886 * gcov.c (struct arc_info): Remove typedef for arc_t.
887 (struct line_info): Likewise.
888 (add_branch_counts): Likewise.
889 (output_branch_count): Likewise.
890 (function_info::~function_info): Likewise.
892 (output_intermediate_line): Likewise.
893 (read_graph_file): Likewise.
894 (solve_flow_graph): Likewise.
895 (find_exception_blocks): Likewise.
896 (add_line_counts): Likewise.
897 (accumulate_line_info): Likewise.
898 (output_line_details): Likewise.
899 (output_function_details): Likewise.
901 2017-11-13 Martin Liska <mliska@suse.cz>
903 * gcov.c (struct function_info): Remove typedef for function_t.
904 (struct source_info): Likewise.
905 (source_info::get_functions_at_location): Likewise.
906 (solve_flow_graph): Likewise.
907 (find_exception_blocks): Likewise.
908 (add_line_counts): Likewise.
909 (output_intermediate_file): Likewise.
910 (process_file): Likewise.
911 (generate_results): Likewise.
912 (release_structures): Likewise.
913 (read_graph_file): Likewise.
914 (read_count_file): Likewise.
915 (accumulate_line_counts): Likewise.
916 (output_lines): Likewise.
918 2017-11-13 Martin Liska <mliska@suse.cz>
920 * gcov.c (function_info::function_info): Remove num_counts
921 and add vector<gcov_type>.
922 (function_info::~function_info): Use the vector.
923 (process_file): Likewise.
924 (read_graph_file): Likewise.
925 (read_count_file): Likewise.
926 (solve_flow_graph): Likewise.
928 2017-11-13 Martin Liska <mliska@suse.cz>
930 * gcov.c (function_info::is_artificial): New function.
931 (process_file): Erase all artificial early.
932 (generate_results): Skip as all artificial are already
935 2017-11-13 Martin Liska <mliska@suse.cz>
937 * gcov.c (read_graph_file): Store to global vector of functions.
938 (read_count_file): Iterate the vector.
939 (process_file): Likewise.
940 (generate_results): Likewise.
941 (release_structures): Likewise.
943 2017-11-13 Jakub Jelinek <jakub@redhat.com>
945 PR tree-optimization/82954
946 * gimple-ssa-store-merging.c
947 (imm_store_chain_info::coalesce_immediate_stores): If
948 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
950 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
952 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
953 Upddate call to ENDIAN_LANE_N.
954 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
955 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
956 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
957 and use aarch64_endian_lane_rtx.
958 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
960 2017-11-12 Tom de Vries <tom@codesourcery.com>
962 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
964 2017-11-12 Tom de Vries <tom@codesourcery.com>
966 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
968 2017-11-12 Tom de Vries <tom@codesourcery.com>
970 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
972 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
973 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
974 * defaults.h (ASM_OUTPUT_LABELREF): Same.
976 2017-11-11 Martin Sebor <msebor@redhat.com>
979 * doc/extend.texi (attribute nonstring): Remove spurious argument.
982 * prefic.c (translate_name): Replace strncpy with memcpy to
983 avoid -Wstringop-truncation.
985 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
987 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
989 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
991 * predict.c (maybe_hot_frequency_p): Do not use cfun.
993 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
995 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
998 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1000 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1001 enable generating XXBRH if the value is in a vector register.
1002 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1003 value is in a vector register.
1004 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1005 register to register bswap64's instead of doing the GPR sequence
1006 used on previous machines.
1007 (bswapdi2_xxbrd): New insn.
1008 (bswapdi2_reg): Disallow on ISA 3.0.
1009 (register to register bswap64 splitter): Do not split the insn on
1010 ISA 3.0 systems that use XXBRD.
1012 2017-11-10 Martin Sebor <msebor@redhat.com>
1015 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1017 (find_subframework_file): Same.
1019 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1021 * auto-profile.c (afdo_indirect_call): Drop frequency.
1022 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1023 (cgraph_node::create_edge): Drop frequency argument.
1024 (cgraph_node::create_indirect_edge): Drop frequency argument.
1025 (cgraph_edge::make_speculative): Drop frequency arguments.
1026 (cgraph_edge::resolve_speculation): Do not update frequencies
1027 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1028 (cgraph_node::dump): Check consistency in IPA mode.
1029 (cgraph_edge::maybe_hot_p): Use IPA counter.
1030 (cgraph_edge::verify_count_and_frequency): Rename to ...
1031 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1032 (cgraph_node::verify_node): Update.
1033 * cgraph.h (struct cgraph_edge): Drop frequency.
1034 (cgraph_edge::frequency): New function.
1035 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1037 (cgraph_edge::rebuild_edges): Likewise.
1038 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1039 (duplicate_thunk_for_node): Do not pass frequency.
1040 (cgraph_node::create_clone): Scale only counts.
1041 (cgraph_node::create_virtual_clone): Do not pass frequency.
1042 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1043 (cgraph_node::create_version_clone): Do not pass frequency.
1044 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1045 (cgraph_node::expand_thunk): Do not pass frequency.
1046 (cgraph_node::create_wrapper): Do not pass frequency.
1047 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1049 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1050 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1051 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1052 (gather_caller_stats): Use frequency function.
1053 (ipcp_cloning_candidate_p): Use frequency function.
1054 (ipcp_propagate_stage): Use frequency function.
1055 (get_info_about_necessary_edges): Use frequency function.
1056 (update_profiling_info): Update only IPA profile.
1057 (update_specialized_profile): Use frequency functoin.
1058 (perhaps_add_new_callers): Update only IPA profile.
1059 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1060 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1061 (dump_ipa_call_summary): Use frequency function.
1062 (estimate_edge_size_and_time): Use frequency function.
1063 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1064 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1065 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1066 (update_noncloned_counts): ... ths one; scale counts only.
1067 (clone_inlined_nodes): Do not scale frequency.
1068 (inline_call): Do not pass frequency.
1069 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1070 (compute_inlined_call_time): Use IPA profile.
1071 (want_inline_small_function_p): Use IPA profile.
1072 (want_inline_self_recursive_call_p): Use IPA profile.
1073 (edge_badness): Use IPA profile.
1074 (lookup_recursive_calls): Use IPA profile.
1075 (recursive_inlining): Do not pass frequency.
1076 (resolve_noninline_speculation): Do not update frequency.
1077 (inline_small_functions): Collect max of IPA profile.
1078 (dump_overall_stats): Dump IPA porfile.
1079 (dump_inline_stats): Dump IPA porfile.
1080 (ipa_inline): Collect IPA stats.
1081 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1082 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1083 (ipa_propagate_frequency): Use frequency function.
1084 (ipa_profile): Cleanup.
1085 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1086 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1087 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1088 (input_node): Do not stream frequency.
1089 (input_edge): Do not stream frequency.
1090 (merge_profile_summaries): Scale only IPA profiles.
1091 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1092 * predict.c (drop_profile): Do not recompute frequency.
1093 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1094 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1095 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1096 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1097 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1098 * tree-inline.c (copy_bb): Do not scale frequency.
1099 (expand_call_inline): Do not scale frequency.
1100 (tree_function_versioning): Do not scale frequency.
1101 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1103 2017-11-10 Julia Koval <julia.koval@intel.com>
1105 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1106 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1107 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1108 (_mm256_maskz_gf2p8affine_epi64_epi8)
1109 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1110 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1111 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1112 (__builtin_ia32_vgf2p8affineqb_v32qi)
1113 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1114 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1116 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1120 (arm_option_override): Refactor.
1121 (arm_option_reconfigure_globals): New.
1122 (arm_options_perform_arch_sanity_checks): New.
1123 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1125 (arm_options_perform_arch_sanity_checks): Likewise
1127 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1129 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1130 (power9-qpmul): New.
1131 * rs6000/rs6000.md ("type" attr): Add qmul.
1132 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1133 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1134 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1136 2017-11-10 Martin Sebor <msebor@redhat.com>
1139 * builtins.c (compute_objsize): Handle arrays that
1140 compute_builtin_object_size likes to fail for. Make extern.
1141 * builtins.h (compute_objsize): Declare.
1142 (check_strncpy_sizes): New function.
1143 (expand_builtin_strncpy): Call check_strncpy_sizes.
1144 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1145 -Wstringop-truncation.
1146 (gimple_fold_builtin_strncat): Same.
1147 * gimple.c (gimple_build_call_from_tree): Set call location.
1148 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1149 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1150 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1151 (handle_builtin_strlen): Use strlen_to_stridx.
1152 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1154 Use strlen_to_stridx.
1155 (pass_strlen::execute): Release strlen_to_stridx.
1156 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1157 (-Wstringop-truncation): Document new option.
1159 2017-11-10 Martin Liska <mliska@suse.cz>
1161 PR gcov-profile/82702
1162 * gcov.c (main): Handle intermediate files in a different
1164 (get_gcov_intermediate_filename): New function.
1165 (output_gcov_file): Remove support of intermediate files.
1166 (generate_results): Allocate intermediate file.
1167 (release_structures): Clean-up properly fn_end.
1168 (output_intermediate_file): Start iterating with line 1.
1170 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1172 PR tree-optimization/82929
1173 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1174 ops_swapped_p non-static data member.
1175 (store_immediate_info::store_immediate_info): Clear it.
1176 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1177 ops set ops_swapped_p.
1178 (count_multiple_uses): Handle ops_swapped_p.
1180 2017-11-10 Martin Liska <mliska@suse.cz>
1182 * coverage.c (coverage_init): Stream information about
1183 support of has_unexecuted_blocks.
1184 * doc/gcov.texi: Document that.
1185 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1186 * gcov.c (read_graph_file): Likewise.
1187 (output_line_beginning): Fix a small issue with
1190 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1192 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1193 reference of trivial component.
1195 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1198 * gimple-ssa-store-merging.c
1199 (pass_store_merging::terminate_all_aliasing_chains): For
1200 gimple_store_p stmts also call refs_output_dependent_p.
1202 PR rtl-optimization/82913
1203 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1206 2017-11-09 Jeff Law <law@redhat.com>
1208 * vr-values.h: New file with vr_values class.
1209 * tree-vrp.c: Include vr-values.h
1210 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1211 data objects into the vr_values class.
1212 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1213 (get_value_range): Make it a member function within vr_values class.
1214 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1215 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1216 (op_with_constant_singleton_value_range): Likewise.
1217 (extract_range_for_var_from_comparison_expr): Likewise.
1218 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1219 (extract_range_from_binary_expr): Likewise.
1220 (extract_range_from_unary_expr): Likewise.
1221 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1222 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1223 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1224 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1225 (compare_name_with_value, compare_names): Likewise.
1226 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1227 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1228 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1229 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1230 (extract_range_from_phi_node): Likewise.
1231 (simplify_truth_ops_using_ranges): Likewise.
1232 (simplify_div_or_mod_using_ranges): Likewise.
1233 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1234 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1235 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1236 (simplify_float_conversion_using_ranges): Likewise.
1237 (simplify_internal_call_using_ranges): Likewise.
1238 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1239 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1240 poitner via x_vr_values for calls into gimple folder.
1241 (vrp_initialize_lattice): Make this the vr_values ctor.
1242 (vrp_free_lattice): Make this the vr_values dtor.
1243 (set_vr_value): New function.
1244 (class vrp_prop): Add vr_values data member. Add various member
1245 functions as well as member functions that delegate to vr_values.
1246 (check_array_ref): Make a member function within vrp_prop class.
1247 (search_for_addr_array, vrp_initialize): Likewise.
1248 (vrp_finalize): Likewise. Revamp to avoid direct access to
1249 vr_value, values_propagated, etc.
1250 (check_array_bounds): Extract vrp_prop class instance pointer from
1251 walk info structure. Use it to call member functions.
1252 (check_all_array_refs): Make a member function within vrp_prop class.
1253 Smuggle class instance pointer via walk info structure.
1254 (x_vr_values): New local static.
1255 (vrp_valueize): Use x_vr_values to get class instance.
1256 (vr_valueize_1): Likewise.
1257 (class vrp_folder): Add vr_values data member. Add various member
1258 functions as well as member functions that delegate to vr_values.
1259 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1260 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1261 class instance from vr_values. Use it to call member functions.
1262 (vrp_dom_walker): Add vr_values data member.
1263 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1264 instance via x_vr_values.
1265 (identify_jump_threads): Accept vr_values as argument. Store
1266 it into the walker structure.
1267 (evrp_dom_walker): Add vr_values class data member. Add various
1269 (evrp_dom_walker::try_find_new_range): Use vr_values data
1270 member to access the memory allocator.
1271 (evrp_dom_walker::before_dom_children): Store vr_values class
1272 instance into the vrp_folder class.
1273 (evrp_dom_walker::push_value_range): Rework to avoid direct
1274 access to num_vr_values and vr_value.
1275 (evrp_dom_walker::pop_value_range): Likewise.
1276 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1277 Use vr_values to get to dump_all_value_ranges member function.
1278 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1279 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1280 Pass vr_values class instance down to identify_jump_threads.
1281 Remove call to vrp_free_lattice.
1282 (debug_all_value_ranges): Remove.
1284 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1285 (vrp_folder): Likewise.
1287 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1288 Get it from the existing bitmap instead.
1289 (vrp_intersect_ranges_1): Likewise.
1291 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1293 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1295 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1296 set bit_not_p to it.
1297 (imm_store_chain_info::coalesce_immediate_stores): Break group
1298 if bit_not_p is different.
1299 (count_multiple_uses, split_group,
1300 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1301 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1302 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1303 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1304 to store_immediate_info ctor.
1306 2017-11-09 Jim Wilson <jimw@sifive.com>
1308 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1309 (scan_prog_file): Likewise.
1311 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1313 * bb-reorder.c (max_entry_frequency): Remove.
1314 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1315 connect_traces, push_to_next_round_p): Remove prototypes.
1316 (find_traces_1_round): Use counts only.
1317 (push_to_next_round_p): Likewise.
1318 (find_traces): Likewise.
1319 (rotate_loop): Likewise.
1320 (find_traces_1_round): Likewise.
1321 (connect_traces): Likewise.
1322 (edge_order): Likewise.
1324 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1326 * config/arm/arm.c (output_return_instruction): Add comments to
1327 indicate requirement for cmse_nonsecure_entry return to account
1328 for the size of clearing instruction output here.
1329 (thumb_exit): Likewise.
1330 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1331 return in hardfloat mode.
1333 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1335 * config/rs6000/rs6000.c (machine_function): Add a bool,
1336 "toc_is_wrapped_separately".
1337 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1338 if it wasn't explicitly set or unset, we are optimizing for speed, and
1339 doing separate shrink-wrapping.
1340 (rs6000_get_separate_components): Enable the TOC component if
1341 saving the TOC register in the prologue.
1342 (rs6000_components_for_bb): Handle the TOC component.
1343 (rs6000_emit_prologue_components): Store the TOC register where needed.
1344 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1345 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1347 2017-11-09 Martin Jambor <mjambor@suse.cz>
1349 * ipa-param-manipulation.c: New file.
1350 * ipa-param-manipulation.h: Likewise.
1351 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1352 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1353 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1354 (ipa_parm_adjustment): Likewise.
1355 (ipa_parm_adjustment_vec): Likewise.
1356 (ipa_get_vector_of_formal_parms): Moved declaration to
1357 ipa-param-manipulation.h.
1358 (ipa_get_vector_of_formal_parm_types): Likewise.
1359 (ipa_modify_formal_parameters): Likewise.
1360 (ipa_modify_call_arguments): Likewise.
1361 (ipa_combine_adjustments): Likewise.
1362 (ipa_dump_param_adjustments): Likewise.
1363 (ipa_modify_expr): Likewise.
1364 (ipa_get_adjustment_candidate): Likewise.
1365 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1366 ipa-param-manipulation.c.
1367 (ipa_get_vector_of_formal_parm_types): Likewise.
1368 (ipa_modify_formal_parameters): Likewise.
1369 (ipa_modify_call_arguments): Likewise.
1370 (ipa_modify_expr): Likewise.
1371 (get_ssa_base_param): Likewise.
1372 (ipa_get_adjustment_candidate): Likewise.
1373 (index_in_adjustments_multiple_times_p): Likewise.
1374 (ipa_combine_adjustments): Likewise.
1375 (ipa_dump_param_adjustments): Likewise.
1376 * tree-sra.c: Also include ipa-param-manipulation.h
1377 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1380 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1381 Alan Hayward <alan.hayward@arm.com>
1382 David Sherwood <david.sherwood@arm.com>
1384 * doc/sourcebuild.texi (vect_masked_store): Document.
1386 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1392 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1393 Alan Hayward <alan.hayward@arm.com>
1394 David Sherwood <david.sherwood@arm.com>
1396 * doc/sourcebuild.texi (vect_variable_length): Document.
1398 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1399 Alan Hayward <alan.hayward@arm.com>
1400 David Sherwood <david.sherwood@arm.com>
1402 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1404 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1410 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1415 previously undocumented selectors.
1416 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1418 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1420 * doc/rtl.texi (const_vector): Say that elements can be
1421 const_wide_ints too.
1422 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1423 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1424 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1425 * optabs.c (expand_vector_broadcast): Likewise.
1427 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1428 Alan Hayward <alan.hayward@arm.com>
1429 David Sherwood <david.sherwood@arm.com>
1431 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1432 scaled index even if the unscaled address was invalid.
1433 Don't increase the complexity of using a scale in that case.
1435 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1436 Alan Hayward <alan.hayward@arm.com>
1437 David Sherwood <david.sherwood@arm.com>
1439 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1440 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1441 UNITS_PER_WORD bytes.
1442 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1443 based on REGMODE_NATURAL_SIZE of the inner mode.
1444 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1445 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1446 * expmed.c (lowpart_bit_field_p): Divide the value up into
1447 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1448 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1449 whether something is likely to occupy more than one register.
1451 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1454 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1455 profile_count::adjust_for_ipa_scaling.
1456 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1457 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1459 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1461 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1463 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1464 (split_group): Add total_orig and total_new arguments, estimate the
1465 number of statements related to the store group without store merging
1466 and with store merging.
1467 (imm_store_chain_info::output_merged_store): Adjust split_group
1468 callers, punt if estimated number of statements with store merging
1469 is not smaller than estimated number of statements without it.
1471 (handled_load): Remove has_single_use checks.
1472 (pass_store_merging::process_store): Likewise.
1474 2017-11-09 Richard Biener <rguenther@suse.de>
1476 PR tree-optimization/82902
1477 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1479 2017-11-09 Martin Liska <mliska@suse.cz>
1482 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1485 2017-11-09 Martin Liska <mliska@suse.cz>
1487 PR tree-optimization/82669
1488 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1490 2017-11-09 Martin Liska <mliska@suse.cz>
1492 PR gcov-profile/48463
1493 * coverage.c (coverage_begin_function): Output also end locus
1494 of a function and information whether the function is
1496 * gcov-dump.c (tag_function): Parse and print the information.
1497 * gcov.c (INCLUDE_MAP): Add include.
1498 (INCLUDE_SET): Likewise.
1499 (struct line_info): Move earlier in the source file because
1500 of vector<line_info> in function_info structure.
1501 (line_info::line_info): Likewise.
1502 (line_info::has_block): Likewise.
1503 (struct source_info): Add new member index.
1504 (source_info::get_functions_at_location): New function.
1505 (function_info::group_line_p): New function.
1506 (output_intermediate_line): New function.
1507 (output_intermediate_file): Use the mentioned function.
1508 (struct function_start): New.
1509 (struct function_start_pair_hash): Likewise.
1510 (process_file): Add code that identifies group functions.
1511 Assign lines either to global or function scope.
1512 (generate_results): Skip artificial functions.
1513 (find_source): Assign index for each source file.
1514 (read_graph_file): Read new flag artificial and end_line.
1515 (add_line_counts): Assign it either to global of function scope.
1516 (accumulate_line_counts): Isolate core of the function to
1517 accumulate_line_info and call it for both function and global
1519 (accumulate_line_info): New function.
1520 (output_line_beginning): Fix GNU coding style.
1521 (print_source_line): New function.
1522 (output_line_details): Likewise.
1523 (output_function_details): Likewise.
1524 (output_lines): Iterate both source (global) scope and function
1526 (struct function_line_start_cmp): New class.
1527 * doc/gcov.texi: Reflect changes in documentation.
1529 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1532 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1533 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1534 and similarly for not instead of neg.
1536 2017-11-08 Andi Kleen <ak@linux.intel.com>
1538 * config/i386/i386.opt: Add -mforce-indirect-call.
1539 * config/i386/predicates.md: Check for flag_force_indirect_call.
1540 * doc/invoke.texi: Document -mforce-indirect-call
1542 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1544 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1546 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1547 config/riscv/riscv.c (predict.h): New include.
1548 (riscv_slow_unaligned_access_p): No longer static.
1549 (riscv_block_move_straight): Add require.
1550 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1552 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1555 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1556 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1557 nonimmediate_operand predicate for operand 1 instead of
1560 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1562 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1564 * config/aarch64/constraints.md (Uml): New constraint.
1565 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1568 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1570 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1571 of two vec_duplicates into a vec_concat.
1573 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1575 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1576 Simplify vec_merge of vec_duplicate and vec_concat.
1577 * config/aarch64/constraints.md (Utq): New constraint.
1578 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1581 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1583 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1584 Simplify vec_merge of vec_duplicate and const_vector.
1585 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1587 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1588 mode iterator. Update predicate on operand 1 to
1589 handle non-const_vec constants. Delete constraints.
1590 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1592 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1594 PR tree-optimization/78821
1595 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1597 (store_operand_info::store_operand_info): Initialize it to false.
1598 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1599 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1600 store in the group, and if chain_info is non-NULL, to ignore altogether
1602 (compatible_load_p): Fail if bit_not_p does not match.
1603 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1604 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1605 (pass_store_merging::process_store): Adjust
1606 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1607 call terminate_all_aliasing_chains newly when adding a store into
1608 a chain with non-NULL chain_info.
1610 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1612 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1614 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1616 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1618 (aarch64_layout_frame): Initialise emit_frame_chain.
1619 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1620 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1622 2017-11-08 Martin Liska <mliska@suse.cz>
1624 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1625 of gimple_call_internal_p.
1627 2017-11-07 Tom de Vries <tom@codesourcery.com>
1629 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1631 2017-11-07 Tom de Vries <tom@codesourcery.com>
1633 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1636 2017-11-08 Martin Liska <mliska@suse.cz>
1639 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1641 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1643 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1645 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1648 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1649 (*cmp<mode>_ccz_1): New insn with $k alternative.
1652 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1653 RTX_COMM_COMPARE as commutative as well.
1654 (ix86_binary_operator_ok): Formatting fix.
1655 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1656 *<code><mode>3<mask_name><round_saeonly_name>,
1657 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1658 *<s>mul<mode>3_highpart<mask_name>,
1659 *vec_widen_umult_even_v16si<mask_name>,
1660 *vec_widen_umult_even_v8si<mask_name>,
1661 *vec_widen_umult_even_v4si<mask_name>,
1662 *vec_widen_smult_even_v16si<mask_name>,
1663 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1664 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1665 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1666 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1667 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1668 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1669 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1670 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1671 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1672 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1673 ix86_binary_operator_ok. Formatting fixes.
1674 (*<plusminus_insn><mode>3<mask_name><round_name>,
1675 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1678 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1680 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1681 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1682 eq and ne if TARGET_ISEL.
1683 (cmp): New code_iterator.
1684 (UNS, UNSU_, UNSIK): New code_attrs.
1685 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1686 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1688 ("eq<mode>3"): ... this.
1689 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1691 ("ne<mode>3"): ... this.
1693 2017-11-07 Julia Koval <julia.koval@intel.com>
1696 * common/config/i386/i386-common.c
1697 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1698 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1699 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1700 * config/i386/i386.opt: Ditto.
1701 * config/i386/i386.c (ix86_target_string): Ditto.
1702 (ix86_option_override_internal): Ditto.
1703 (ix86_init_mpx_builtins): Move MPX to args2.
1704 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1705 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1706 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1707 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1708 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1709 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1710 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1712 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1715 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1716 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1717 (zero-extendsidi peephole2): Remove peephole.
1719 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1722 * stor-layout.c: Include attribs.h.
1723 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1724 explicit lookup of "aligned" attribute.
1726 2017-11-07 Andrew Waterman <andrew@sifive.com>
1728 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1729 (riscv_expand_block_move): Likewise.
1730 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1732 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1733 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1734 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1735 (riscv_adjust_block_mem): Likewise.
1736 (riscv_block_move_loop): Likewise.
1737 (riscv_expand_block_move): Likewise.
1738 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1740 2017-11-07 Michael Clark <michaeljclark@mac.com>
1742 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1743 (MUSL_DYNAMIC_LINKER): Likewise.
1745 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1747 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1748 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1751 2017-11-07 Richard Biener <rguenther@suse.de>
1753 * match.pd: Fix build.
1755 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1756 Jackson Woodruff <jackson.woodruff@arm.com>
1758 PR tree-optimization/71026
1759 * match.pd: Canonicalize negate in division.
1761 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1764 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1766 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1768 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1769 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1771 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1773 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1774 non-scalar integral types.
1775 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1776 (-(A-B), -(~A)): New transformations.
1778 2017-11-07 Tom de Vries <tom@codesourcery.com>
1780 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1781 semicolon after "do {} while (0)".
1782 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1783 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1784 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1785 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1786 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1788 2017-11-07 Tom de Vries <tom@codesourcery.com>
1790 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1791 after "do {} while (0)".
1792 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1793 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1794 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1795 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1796 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1798 2017-11-07 Tom de Vries <tom@codesourcery.com>
1801 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1803 (arm_rtx_costs_internal): Add missing semicolon after
1804 HANDLE_NARROW_SHIFT_ARITH call.
1806 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1808 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1809 disable isel if it was not set explicitly.
1811 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1813 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1815 (add_type_attribute) likewise.
1817 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1819 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1820 of struct ix86_frame.
1821 (ix86_initial_elimination_offset): Likewise.
1822 (ix86_expand_split_stack_prologue): Likewise.
1824 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1826 * tree-vrp.h (enum value_range_type): Update stale comment.
1828 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1829 Alan Hayward <alan.hayward@arm.com>
1830 David Sherwood <david.sherwood@arm.com>
1832 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1833 (aarch64_expand_vec_perm_const): Take the number of units too.
1834 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1835 (aarch64_expand_vec_perm_const): Likewise.
1836 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1837 (vec_perm<mode>): Update accordingly.
1839 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1840 Alan Hayward <alan.hayward@arm.com>
1841 David Sherwood <david.sherwood@arm.com>
1843 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1844 Take the number of units too.
1845 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1846 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1847 but check for a vector mode before rather than after the call.
1848 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1849 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1850 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1851 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1852 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1853 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1854 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1855 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1856 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1857 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1858 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1859 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1860 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1861 (aarch64_sqdmull2_n<mode>): Update accordingly.
1863 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1864 Alan Hayward <alan.hayward@arm.com>
1865 David Sherwood <david.sherwood@arm.com>
1867 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1868 the number of units too.
1869 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1870 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1871 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1872 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1873 (vec_store_lanesxi<mode>): Update accordingly.
1875 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1876 Alan Hayward <alan.hayward@arm.com>
1877 David Sherwood <david.sherwood@arm.com>
1879 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1880 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1881 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1882 of units rather than the mode.
1883 * config/aarch64/iterators.md (nunits): New mode attribute.
1884 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1885 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1886 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1887 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1888 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1889 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1890 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1891 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1892 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1893 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1894 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1895 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1896 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1897 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1898 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1899 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1900 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1901 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1902 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1903 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1904 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1905 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1906 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1907 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1908 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1909 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1910 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1911 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1912 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1913 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1914 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1915 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1916 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1917 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1918 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1919 (aarch64_simd_vec_setv2di): Likewise.
1921 2017-11-06 Carl Love <cel@us.ibm.com>
1923 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1925 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1926 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1928 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1930 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1931 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1932 Add power 8 macro expansions.
1933 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1934 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1935 power 8 instructions. (VSX_XXBR): Add iterator.
1937 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1939 * config/arm/arm.md (predicable_short_it): Change default to "no",
1940 improve documentation, remove uses that are identical to the default.
1941 (enabled_for_depr_it): Rename to enabled_for_short_it.
1942 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1943 Remove default uses.
1944 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1945 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1946 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1947 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1949 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1952 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1953 float128 helper macros, which are no longer used after deleting
1954 the old 'q' built-in functions, and moving the round to odd
1955 built-in functions to being special built-in functions.
1956 (BU_FLOAT128_2): Likewise.
1957 (BU_FLOAT128_1_HW): Likewise.
1958 (BU_FLOAT128_2_HW): Likewise.
1959 (BU_FLOAT128_3_HW): Likewise.
1960 (FABSQ): Delete old 'q' built-in functions.
1961 (COPYSIGNQ): Likewise.
1962 (SQRTF128_ODD): Move round to odd built-in functions to be
1963 special built-in functions, so that we can handle
1964 -mabi=ieeelongdouble.
1965 (TRUNCF128_ODD): Likewise.
1966 (ADDF128_ODD): Likewise.
1967 (SUBF128_ODD): Likewise.
1968 (MULF128_ODD): Likewise.
1969 (DIVF128_ODD): Likewise.
1970 (FMAF128_ODD): Likewise.
1971 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1972 built-in names to 'f128'.
1973 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1974 old 'q' built-in functions, as the machine independent code for
1975 'f128' built-in functions handles this.
1976 (rs6000_expand_builtin): Add expansion for float128 round to odd
1977 functions, keying off on -mabi=ieeelongdouble of whether to use
1978 the KFmode or TFmode variant.
1979 (rs6000_init_builtins): Initialize the _Float128 round to odd
1981 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1982 _Float128 'q' built-in functions are now mapped into the new
1983 'f128' built-in functions.
1985 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1987 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1988 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1990 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1993 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1994 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1995 * toplev.c: Include "ipa-fnsummary.h".
1996 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1998 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2000 PR tree-optimization/82838
2001 * gimple-ssa-store-merging.c
2002 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2003 on a separate gimple_seq which is then appended to seq.
2005 2017-11-06 Jeff Law <law@redhat.com>
2008 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2009 (get_probe_interval): New functions.
2010 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2011 (ix86_adjust_stack_and_probe): Likewise.
2012 (output_adjust_stack_and_probe): Likewise.
2013 (ix86_emit_probe_stack_range): Likewise.
2014 (ix86_expand_prologue): Likewise.
2016 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2018 PR tree-optimization/82816
2019 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2020 if the modes of the two types are the same.
2021 (convert_plusminus_to_widen): Likewise.
2023 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2025 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2027 (usadv16qi): New define_expand.
2028 (usadv8hi): New define_expand.
2030 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2033 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2035 (inline_transform): Likewise.
2036 * predict.c (counts_to_freqs): Remove useless conditional.
2037 * profile-count.h (profile_count::apply_scale): Move sanity check.
2038 * tree-inline.c (copy_bb): Always scale.
2039 (copy_cfg_body): Likewise.
2041 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2044 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2046 (*cmp_ite0): Add enabled_for_depr_it attribute.
2047 (*cmp_ite1): Likewise.
2049 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2051 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2054 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2056 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2057 return the value by pointer.
2058 (extract_range_from_multiplicative_op_1): Update accordingly.
2059 Return as soon as an operation fails.
2061 2017-11-05 Tom de Vries <tom@codesourcery.com>
2064 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2065 (DEF_SANITIZER_BUILTIN): ... here.
2066 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2067 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2069 2017-11-05 Tom de Vries <tom@codesourcery.com>
2072 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2074 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2075 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2076 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2078 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2079 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2080 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2082 2017-11-05 Tom de Vries <tom@codesourcery.com>
2085 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2088 2017-11-04 Michael Clark <michaeljclark@mac.com>
2090 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2091 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2093 (*addsi3_extended): Likewise.
2094 (*addsi3_extended2): Likewise.
2095 (<optab>si3): Likewise.
2096 (<optab>di3): Likewise.
2097 (<optab><mode>3): Likewise.
2098 (<*optabe>si3_internal): Likewise.
2099 (zero_extendqi<SUPERQI:mode>2): Likewise.
2100 (*add<mode>hi3): Likewise.
2101 (*xor<mode>hi3): Likewise.
2102 (<optab>di3): Likewise.
2103 (*<optab>si3_extend): Likewise.
2104 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2105 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2106 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2108 2017-11-04 Andrew Waterman <andrew@sifive.com>
2110 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2111 TARGET_STRICT_ALIGN based upon -mtune argument.
2113 2017-11-04 Andrew Waterman <andrew@sifive.com>
2115 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2117 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2119 config/i386/i386.c (choose_basereg): Use optional scratch
2120 register and add assertion.
2121 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2122 needed, and don't allocate stack.
2123 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2124 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2125 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2127 2017-11-03 Jeff Law <law@redhat.com>
2129 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2130 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2131 to probe at the start of a noreturn function.
2133 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2135 PR tree-optimization/78821
2136 * gimple-ssa-store-merging.c: Update the file comment.
2137 (MAX_STORE_ALIAS_CHECKS): Define.
2138 (struct store_operand_info): New type.
2139 (store_operand_info::store_operand_info): New constructor.
2140 (struct store_immediate_info): Add rhs_code and ops data members.
2141 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2142 and op1r arguments to the ctor, initialize corresponding data members.
2143 (struct merged_store_group): Add load_align_base and load_align
2145 (merged_store_group::merged_store_group): Initialize them.
2146 (merged_store_group::do_merge): Update them.
2147 (merged_store_group::apply_stores): Pick the constant for
2148 encode_tree_to_bitpos from one of the two operands, or skip
2149 encode_tree_to_bitpos if neither operand is a constant.
2150 (class pass_store_merging): Add process_store method decl. Remove
2151 bool argument from terminate_all_aliasing_chains method decl.
2152 (pass_store_merging::terminate_all_aliasing_chains): Remove
2153 var_offset_p argument and corresponding handling.
2154 (stmts_may_clobber_ref_p): New function.
2155 (compatible_load_p): New function.
2156 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2157 if there is overlap and rhs_code is not INTEGER_CST. For
2158 non-overlapping stores terminate group if rhs is not mergeable.
2159 (get_alias_type_for_stmts): Change first argument from
2160 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2161 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2162 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2164 (get_location_for_stmts): Change first argument from
2165 auto_vec<gimple *> & to vec<gimple *> &.
2166 (struct split_store): Remove orig_stmts data member, add orig_stores.
2167 (split_store::split_store): Create orig_stores rather than orig_stmts.
2168 (find_constituent_stmts): Renamed to ...
2169 (find_constituent_stores): ... this. Change second argument from
2170 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2171 to info structures rather than the statements.
2172 (split_group): Rename ALLOW_UNALIGNED argument to
2173 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2174 it. Adjust find_constituent_stores caller.
2175 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2176 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2177 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2178 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2179 (mem_valid_for_store_merging): New function.
2180 (handled_load): New function.
2181 (pass_store_merging::process_store): New method.
2182 (pass_store_merging::execute): Use process_store method. Adjust
2183 terminate_all_aliasing_chains caller.
2185 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2187 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2188 Return true for more constants, symbols and label references.
2189 (aarch64_valid_floating_const): Remove unused function.
2191 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2194 * config/aarch64/aarch64.c (aarch64_layout_frame):
2195 Undo forcing of LR at bottom of frame.
2197 2017-11-03 Jeff Law <law@redhat.com>
2200 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2201 for int_registers_saved.
2203 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2204 extracted from tree-ssa-dom.c.
2205 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2206 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2207 (record_equivalences_from_incoming_edge): Add additional argument
2208 to single_pred_edge_ignoring_loop_edges call.
2209 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2210 (uncprop_dom_walker::before_dom_children): Add additional argument
2211 to single_pred_edge_ignoring_loop_edges call.
2212 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2213 single_pred_edge_ignoring_loop_edges rather than open coding.
2214 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2216 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2218 * match.pd (-(-A)): Rewrite.
2220 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2222 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2223 (rs6000_emit_int_cmove): New declaration.
2224 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2225 (rs6000_emit_sISEL): Delete.
2226 (rs6000_emit_int_cmove): Make non-static.
2227 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2228 instead of rs6000_emit_sISEL.
2230 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2232 * asan.c (create_cond_insert_point): Maintain profile.
2233 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2235 * basic-block.h (struct basic_block_def): Remove frequency.
2236 (EDGE_FREQUENCY): Use to_frequency
2237 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2239 (find_traces): Update to use to_frequency.
2240 (find_traces_1_round): Likewise; use only IPA counts.
2241 (bb_to_key): Likewise.
2242 (connect_traces): Use IPA counts only.
2243 (copy_bb_p): Update to use to_frequency.
2244 (fix_up_crossing_landing_pad): Likewise.
2245 (sanitize_hot_paths): Likewise.
2246 * bt-load.c (basic_block_freq): Likewise.
2247 * cfg.c (init_flow): Set count_max to uninitialized.
2248 (check_bb_profile): Remove frequencies; check counts.
2249 (dump_bb_info): Do not dump frequencies.
2250 (update_bb_profile_for_threading): Update counts only.
2251 (scale_bbs_frequencies_int): Likewise.
2252 (MAX_SAFE_MULTIPLIER): Remove.
2253 (scale_bbs_frequencies_gcov_type): Update counts only.
2254 (scale_bbs_frequencies_profile_count): Update counts only.
2255 (scale_bbs_frequencies): Update counts only.
2256 * cfg.h (struct control_flow_graph): Add count-max.
2257 (update_bb_profile_for_threading): Update prototype.
2258 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2259 (find_many_sub_basic_blocks): Likewise.
2260 * cfgcleanup.c (try_forward_edges): Likewise.
2261 (try_crossjump_to_edge): Likewise.
2262 * cfgexpand.c (expand_gimple_cond): Likewise.
2263 (expand_gimple_tailcall): Likewise.
2264 (construct_init_block): Likewise.
2265 (construct_exit_block): Likewise.
2266 * cfghooks.c (verify_flow_info): Check consistency of counts.
2267 (dump_bb_for_graph): Do not dump frequencies.
2268 (split_block_1): Do not update frequencies.
2269 (split_edge): Do not update frequencies.
2270 (make_forwarder_block): Do not update frequencies.
2271 (duplicate_block): Do not update frequencies.
2272 (account_profile_record): Do not update frequencies.
2273 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2274 for global heuristics.
2275 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2276 (expected_loop_iterations_unbounded): Use counts only.
2277 * cfgloopmanip.c (scale_loop_profile): Simplify.
2278 (create_empty_loop_on_edge): Simplify
2280 (duplicate_loop_to_header_edge): Simplify
2281 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2282 (update_br_prob_note): Take care of removing note when profile
2284 (relink_block_chain): Do not dump frequency.
2285 (rtl_account_profile_record): Use to_frequency.
2286 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2287 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2288 (cgraph_update_edges_for_call_stmt_node): Likewise.
2289 (cgraph_edge::verify_count_and_frequency): Update.
2290 (cgraph_node::verify_node): Temporarily disable frequency verification.
2291 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2292 to_cgraph_frequency.
2293 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2294 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2296 (cgraph_node::expand_thunk): Update profile.
2297 * except.c (dw2_build_landing_pads): Do not update frequency.
2298 * final.c (compute_alignments): Use to_frequency.
2299 (dump_basic_block_info): Do not dump frequency.
2300 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2301 (dump_gimple_bb_header): Do not dump frequency.
2302 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2304 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2305 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2306 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2307 (init_before_recovery): Do not update frequency.
2308 (sched_create_recovery_edges): Do not update frequency.
2309 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2310 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2311 (ipa_cp_c_finalize): Set max_count to uninitialized.
2312 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2313 (param_change_prob): Use counts.
2314 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2316 * ipa-split.c (consider_split): Use to_frequency.
2317 (split_function): Use to_frequency.
2318 * ira-build.c (loop_compare_func): Likewise.
2319 (mark_loops_for_removal): Likewise.
2320 (mark_all_loops_for_removal): Likewise.
2321 * loop-doloop.c (doloop_modify): Do not update frequency.
2322 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2324 * lto-streamer-in.c (input_function): Update count_max.
2325 * omp-expand.c (expand_omp_taskreg): Update count_max.
2326 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2327 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2328 (maybe_hot_count_p): Use ipa counts only.
2329 (maybe_hot_bb_p): Simplify.
2330 (maybe_hot_edge_p): Simplify.
2331 (probably_never_executed): Do not take frequency argument.
2332 (probably_never_executed_bb_p): Do not pass frequency.
2333 (probably_never_executed_edge_p): Likewise.
2334 (combine_predictions_for_bb): Check that profile is nonzero.
2335 (propagate_freq): Do not set frequency.
2336 (drop_profile): Simplify.
2337 (counts_to_freqs): Simplify.
2338 (expensive_function_p): Use to_frequency.
2339 (propagate_unlikely_bbs_forward): Simplify.
2340 (determine_unlikely_bbs): Simplify.
2341 (estimate_bb_frequencies): Add hack to silence graphite issues.
2342 (compute_function_frequency): Use ipa counts.
2343 (pass_profile::execute): Update.
2344 (rebuild_frequencies): Use counts only.
2345 (force_edge_cold): Use counts only.
2346 * profile-count.c (profile_count::dump): Dump new count types.
2347 (profile_count::differs_from_p): Check compatiblity.
2348 (profile_count::to_frequency): New function.
2349 (profile_count::to_cgraph_frequency): New function.
2350 * profile-count.h (struct function): Declare.
2351 (enum profile_quality): Add profile_guessed_local and
2352 profile_guessed_global0.
2353 (class profile_proability): Decrease number of bits to 29;
2354 update from_reg_br_prob_note and to_reg_br_prob_note.
2355 (class profile_count: Update comment; decrease number of bits
2356 to 61. Check compatibility.
2357 (profile_count::compatible_p): New private member function.
2358 (profile_count::ipa_p): New member function.
2359 (profile_count::operator<): Handle global zero correctly.
2360 (profile_count::operator>): Handle global zero correctly.
2361 (profile_count::operator<=): Handle global zero correctly.
2362 (profile_count::operator>=): Handle global zero correctly.
2363 (profile_count::nonzero_p): New member function.
2364 (profile_count::force_nonzero): New member function.
2365 (profile_count::max): New member function.
2366 (profile_count::apply_scale): Handle IPA scalling.
2367 (profile_count::guessed_local): New member function.
2368 (profile_count::global0): New member function.
2369 (profile_count::ipa): New member function.
2370 (profile_count::to_frequency): Declare.
2371 (profile_count::to_cgraph_frequency): Declare.
2372 * profile.c (OVERLAP_BASE): Delete.
2373 (compute_frequency_overlap): Delete.
2374 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2375 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2376 * sched-ebb.c (rank): Use counts only.
2377 * shrink-wrap.c (handle_simple_exit): Use counts only.
2378 (try_shrink_wrapping): Use counts only.
2379 (place_prologue_for_one_component): Use counts only.
2380 * tracer.c (find_best_predecessor): Use to_frequency.
2381 (find_trace): Use to_frequency.
2382 (tail_duplicate): Use to_frequency.
2383 * trans-mem.c (expand_transaction): Do not update frequency.
2384 * tree-call-cdce.c: Do not update frequency.
2385 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2386 (gimple_merge_blocks): Likewise.
2387 (gimple_split_edge): Likewise.
2388 (gimple_duplicate_sese_region): Likewise.
2389 (gimple_duplicate_sese_tail): Likewise.
2390 (move_sese_region_to_fn): Likewise.
2391 (gimple_account_profile_record): Likewise.
2392 (insert_cond_bb): Likewise.
2393 * tree-complex.c (expand_complex_div_wide): Likewise.
2394 * tree-eh.c (lower_resx): Update profile.
2395 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2397 (initialize_cfun): Do not initialize frequencies
2398 (freqs_to_counts): Delete.
2399 (copy_cfg_body): Ignore count parameter.
2400 (copy_body): Update.
2401 (expand_call_inline): Update count_max.
2402 (optimize_inline_calls): Update count_max.
2403 (tree_function_versioning): Update count_max.
2404 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2405 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2407 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2408 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2409 (try_peel_loop): Likewise.
2410 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2412 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2413 (tree_transform_and_unroll_loop): Do not use frequencies
2414 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2415 Use reliable prediction only.
2416 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2417 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2418 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2419 probability scaling.
2420 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2421 not update frequency
2422 (any_remaining_duplicated_blocks): Likewise.
2423 (update_profile): Likewise.
2424 (estimated_freqs_path): Delete.
2425 (freqs_to_counts_path): Delete.
2426 (clear_counts_path): Delete.
2427 (ssa_fix_duplicate_block_edges): Likewise.
2428 (duplicate_thread_path): Likewise.
2429 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2430 * tree-tailcall.c (decrease_profile): Do not update frequency.
2431 (eliminate_tail_call): Likewise.
2432 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2433 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2434 (optimize_mask_stores): Likewise.
2435 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2436 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2437 (ubsan_expand_ptr_ifn): Update profile.
2438 * value-prof.c (gimple_ic): Simplify.
2439 * value-prof.h (gimple_ic): Update prototype.
2440 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2441 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2443 (want_inline_self_recursive_call_p): Likewise.
2444 (resolve_noninline_speculation): Only cummulate defined counts.
2445 (inline_small_functions): Use nonzero_p.
2446 (ipa_inline): Do not access freed node.
2448 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2450 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2451 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2453 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2455 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2456 non-legitimate address.
2458 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2460 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2461 (*lt0_<mode>di, *lt0_<mode>si): New.
2463 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2465 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2466 TARGET_PAIRED_FLOAT.
2468 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2469 Jim Wilson <jim.wilson@linaro.org>
2471 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2472 * config/aarch64/aarch64-tune.md: Regenerated.
2473 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2474 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2476 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2478 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2479 (arc_expand_prologue): Restore blink for millicode.
2480 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2482 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2485 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2486 gen_vec_duplicate after forcing the scalar into a register.
2488 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2490 * combine (try_combine): Print the insns input to try_combine to the
2493 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2496 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2497 Remove second argument from aarch64_process_target_attr call.
2498 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2500 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2502 (aarch64_handle_attr_arch): Remove second argument.
2503 (aarch64_handle_attr_cpu): Ditto.
2504 (aarch64_handle_attr_tune): Ditto.
2505 (aarch64_handle_attr_isa_flags): Ditto.
2506 (aarch64_process_one_target_attr): Ditto.
2507 (aarch64_process_target_attr): Ditto.
2508 (aarch64_option_valid_attribute_p): Remove second argument.
2509 on aarch64_process_target_attr call.
2511 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2513 * diagnostic.c: Include "selftest-diagnostic.h".
2514 (selftest::assert_location_text): New function.
2515 (selftest::test_diagnostic_get_location_text): New function.
2516 (selftest::diagnostic_c_tests): Call it.
2518 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2520 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2521 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2522 (class selftest::test_diagnostic_context): Move to...
2523 * selftest-diagnostic.c: New file.
2524 * selftest-diagnostic.h: New file.
2526 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2528 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2529 offset range for FT32B.
2530 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2531 * config/ft32/ft32.md: Add TARGET_NOPM.
2532 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2533 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2535 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2537 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2539 2017-11-02 Jeff Law <law@redhat.com>
2541 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2542 virtual keyword on FINAL OVERRIDE members.
2544 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2545 virtuals together. Add virtual destructor.
2546 (substitute_and_fold_engine): Similarly.
2548 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2550 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2552 2017-11-02 Richard Biener <rguenther@suse.de>
2554 PR tree-optimization/82795
2555 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2557 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2559 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2561 (gcc_GAS_CHECK_FEATURE): Remove.
2562 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2563 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2564 * configure: Regenerate.
2566 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2568 * config/arc/arc.c (hwloop_optimize): Account for empty
2571 2017-11-02 Richard Biener <rguenther@suse.de>
2574 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2575 Truncate ARRAY_REF index and element size.
2577 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2579 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2581 2017-11-01 Jeff Law <law@redhat.com>
2583 * tree-ssa-ccp.c (ccp_folder): New class derived from
2584 substitute_and_fold_engine.
2585 (ccp_folder::get_value): New member function.
2586 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2587 (ccp_fold_stmt): Remove prototype.
2588 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2589 * tree-ssa-copy.c (copy_folder): New class derived from
2590 substitute_and_fold_engine.
2591 (copy_folder::get_value): Renamed from get_value.
2592 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2593 * tree-vrp.c (vrp_folder): New class derived from
2594 substitute_and_fold_engine.
2595 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2596 (vrp_folder::get_value): New member function.
2597 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2598 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2599 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2600 provide a class interface to folder/substitute routines.
2601 (ssa_prop_fold_stmt_fn): Remove typedef.
2602 (ssa_prop_get_value_fn): Likewise.
2603 (subsitute_and_fold): Remove prototype.
2604 (replace_uses_in): Likewise.
2605 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2606 Renamed from replace_uses_in. Call the virtual member function
2607 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2608 (substitute_and_fold_dom_walker): Remove initialization of
2609 data member entries for calbacks. Add substitute_and_fold_engine
2610 member and initialize it.
2611 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2612 member functions for get_value, replace_phi_args_in c
2613 replace_uses_in, and fold_stmt calls.
2614 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2615 substitute_and_fold. Remove assert. Update ctor call.
2617 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2618 (ssa_prop_visit_phi_fn): Likewise.
2619 (class ssa_propagation_engine): New class to provide an interface
2621 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2623 (ssa_prop_visit_phi): Likewise.
2624 (ssa_propagation_engine::simulate_stmt): Moved into class.
2625 Call visit_phi/visit_stmt from the class rather than via
2626 file scoped static variables.
2627 (ssa_propagation_engine::simulate_block): Moved into class.
2628 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2629 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2630 set file scoped statics for the visit_stmt/visit_phi callbacks.
2631 * tree-complex.c (complex_propagate): New class derived from
2632 ssa_propagation_engine.
2633 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2634 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2635 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2637 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2638 ssa_propagation_engine.
2639 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2640 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2641 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2642 * tree-ssa-copy.c (copy_prop): New class derived from
2643 ssa_propagation_engine.
2644 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2645 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2646 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2647 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2648 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2649 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2650 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2652 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2654 PR rtl-optimization/82778
2655 PR rtl-optimization/82597
2656 * compare-elim.c (struct comparison): Add in_a_setter field.
2657 (find_comparison_dom_walker::before_dom_children): Remove killed
2658 bitmap and df_simulate_find_defs call, instead walk the defs.
2659 Compute last_setter and initialize in_a_setter. Merge definitions
2660 with first initialization for a few variables.
2661 (try_validate_parallel): Use insn_invalid_p instead of
2662 recog_memoized. Return insn rather than just the pattern.
2663 (try_merge_compare): Fix up comment. Don't uselessly test if
2664 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2666 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2669 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2674 aarch64_hard_regno_nregs to get the number of registers
2677 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2678 Alan Hayward <alan.hayward@arm.com>
2679 David Sherwood <david.sherwood@arm.com>
2681 * config/aarch64/constraints.md (Upl): Rename to...
2683 * config/aarch64/aarch64.md
2684 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2687 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2688 Alan Hayward <alan.hayward@arm.com>
2689 David Sherwood <david.sherwood@arm.com>
2691 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2692 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2695 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2700 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2701 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2703 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2704 of a single element.
2705 * config/aarch64/iterators.md: Add a comment above the permute
2706 unspecs to say that they are generated directly by
2707 aarch64_expand_vec_perm_const.
2708 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2710 2017-11-01 Nathan Sidwell <nathan@acm.org>
2712 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2714 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2716 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2717 and medany code models, and describe what they do.
2719 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2721 Revert accidental duplicate:
2723 * combine.c (can_change_dest_mode): Reject changes in
2724 REGMODE_NATURAL_SIZE.
2726 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2728 PR rtl-optimization/64682
2729 PR rtl-optimization/69567
2730 PR rtl-optimization/69737
2731 PR rtl-optimization/82683
2732 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2733 register mentioned in the note, drop the note, unless it came from I3,
2734 in which case it should go to I3 again.
2736 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2738 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2739 and return false if not.
2740 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2742 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2744 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2745 range is known to be empty.
2747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2752 and CONST_VECTOR_NUNITS instead of computing the number of units from
2753 the byte sizes of the vector and element.
2754 (simplify_binary_operation_1): Likewise.
2755 (simplify_const_binary_operation): Likewise.
2756 (simplify_ternary_operation): Likewise.
2758 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2759 Alan Hayward <alan.hayward@arm.com>
2760 David Sherwood <david.sherwood@arm.com>
2762 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2763 (int_mem_offset): ...this new function.
2764 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2765 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2766 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2769 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2770 Alan Hayward <alan.hayward@arm.com>
2771 David Sherwood <david.sherwood@arm.com>
2773 * lower-subreg.c (interesting_mode_p): New function.
2774 (compute_costs, find_decomposable_subregs, decompose_register)
2775 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2776 (resolve_clobber, dump_choices): Use it.
2778 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2779 Alan Hayward <alan.hayward@arm.com>
2780 David Sherwood <david.sherwood@arm.com>
2782 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2784 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * alias.c (find_base_value, find_base_term): Only process integer
2789 truncations. Check the precision rather than the size.
2791 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2792 Alan Hayward <alan.hayward@arm.com>
2793 David Sherwood <david.sherwood@arm.com>
2795 * machmode.h (is_narrower_int_mode): New function
2796 * optabs.c (expand_float, expand_fix): Use it.
2797 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2799 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2800 Alan Hayward <alan.hayward@arm.com>
2801 David Sherwood <david.sherwood@arm.com>
2803 * rtl.h (narrower_subreg_mode): New function.
2804 * ira-color.c (update_costs_from_allocno): Use it.
2806 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2807 Alan Hayward <alan.hayward@arm.com>
2808 David Sherwood <david.sherwood@arm.com>
2810 * optabs-query.h (convert_optab_p): New function, split out from...
2811 (convert_optab_handler): ...here.
2812 (widening_optab_handler): Delete.
2813 (find_widening_optab_handler): Remove permit_non_widening parameter.
2814 (find_widening_optab_handler_and_mode): Likewise. Provide an
2815 override that operates on mode class wrappers.
2816 * optabs-query.c (widening_optab_handler): Delete.
2817 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2818 parameter. Assert that the two modes are the same class and that
2819 the "from" mode is narrower than the "to" mode. Use
2820 convert_optab_handler instead of widening_optab_handler.
2821 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2822 instead of widening_optab_handler.
2823 * expr.c (expand_expr_real_2): Update calls to
2824 find_widening_optab_handler.
2825 * optabs.c (expand_widen_pattern_expr): Likewise.
2826 (expand_binop_directly): Take the insn_code as a parameter.
2827 (expand_binop): Only call find_widening_optab_handler for
2828 conversion optabs; use optab_handler otherwise. Update calls
2829 to find_widening_optab_handler and expand_binop_directly.
2830 Use convert_optab_handler instead of widening_optab_handler.
2831 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2832 find_widening_optab_handler and use scalar_mode rather than
2834 (convert_plusminus_to_widen): Likewise.
2836 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2837 Alan Hayward <alan.hayward@arm.com>
2838 David Sherwood <david.sherwood@arm.com>
2840 * machmode.h (fixed_size_mode): New class.
2841 * rtl.h (get_pool_mode): Return fixed_size_mode.
2842 * gengtype.c (main): Add fixed_size_mode.
2843 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2844 (get_raw_arg_mode): Likewise.
2845 * doc/tm.texi: Regenerate.
2846 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2847 * targhooks.c (default_get_reg_raw_mode): Likewise.
2848 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2849 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2850 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2851 (msp430_get_raw_result_mode): Likewise.
2852 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2853 * dbxout.c (dbxout_parms): Require fixed-size modes.
2854 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2855 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2856 * omp-low.c (lower_oacc_reductions): Likewise.
2857 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2858 (simplify_subreg): Update accordingly.
2859 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2860 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2861 that aren't fixed-size.
2862 (get_pool_mode): Return a fixed_size_mode.
2863 (output_constant_pool_2): Take a fixed_size_mode.
2865 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2866 Alan Hayward <alan.hayward@arm.com>
2867 David Sherwood <david.sherwood@arm.com>
2869 * doc/rtl.texi (vec_series): Document.
2870 (const): Say that the operand can be a vec_series.
2871 * rtl.def (VEC_SERIES): New rtx code.
2872 * rtl.h (const_vec_series_p_1): Declare.
2873 (const_vec_series_p): New function.
2874 * emit-rtl.h (gen_const_vec_series): Declare.
2875 (gen_vec_series): Likewise.
2876 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2877 (gen_vec_series): Likewise.
2878 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2879 * simplify-rtx.c (simplify_unary_operation): Handle negations
2881 (simplify_binary_operation_series): New function.
2882 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2883 (test_vector_ops_series): New function.
2884 (test_vector_ops): Call it.
2885 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2886 gen_const_vec_series.
2887 (altivec_lvsr): Likewise.
2888 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2891 Alan Hayward <alan.hayward@arm.com>
2892 David Sherwood <david.sherwood@arm.com>
2894 * doc/rtl.texi (const): Update description of address constants.
2895 Say that vector constants are allowed too.
2896 * common.md (E, F): Use CONSTANT_P instead of checking for
2898 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2899 checking for CONST_VECTOR.
2900 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2902 * expr.c (expand_expr_real_2): Check for vector modes instead
2903 of checking for CONST_VECTOR.
2904 * rtl.h (const_vec_p): New function.
2905 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2906 (unwrap_const_vec_duplicate): Handle them here too.
2908 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2909 David Malcolm <dmalcolm@redhat.com>
2910 Alan Hayward <alan.hayward@arm.com>
2911 David Sherwood <david.sherwood@arm.com>
2913 * rtl.h (vec_duplicate_p): New function.
2914 * selftest-rtl.c (assert_rtx_eq_at): New function.
2915 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2916 (assert_rtx_eq_at): Declare.
2917 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2918 * selftest-run-tests.c (selftest::run_tests): Call it.
2919 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2920 (simplify_unary_operation_1): Recursively handle vector duplicates.
2921 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2923 (simplify_subreg): Handle subregs of vector duplicates.
2924 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2925 (selftest::simplify_rtx_c_tests): New functions.
2927 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2928 Alan Hayward <alan.hayward@arm.com>
2929 David Sherwood <david.sherwood@arm.com>
2931 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2932 (gen_vec_duplicate): Likewise.
2933 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2935 (gen_const_vector): ...here.
2936 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2937 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2938 whose elements are all equal.
2939 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2940 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2941 (simplify_relational_operation): Likewise.
2942 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2944 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2945 (aarch64_expand_vector_init): Likewise.
2946 * config/arm/arm.c (neon_vdup_constant): Likewise.
2947 (neon_expand_vector_init): Likewise.
2948 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2949 (arm_block_set_unaligned_vect): Likewise.
2950 (arm_block_set_aligned_vect): Likewise.
2951 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2952 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2953 (expand_vec_perm_even_odd_pack): Likewise.
2954 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2955 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2956 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2957 gen_const_vec_duplicate.
2958 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2959 * config/mips/mips.c (mips_gen_const_int_vector): Use
2960 gen_const_vec_duplicate.
2961 (mips_expand_vector_init): Use CONST0_RTX.
2962 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2963 (define_split): Use gen_const_vec_duplicate.
2964 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2965 (define_split): Use gen_const_vec_duplicate.
2966 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2967 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2968 * config/spu/spu.c (spu_const): Likewise.
2970 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2971 Alan Hayward <alan.hayward@arm.com>
2972 David Sherwood <david.sherwood@arm.com>
2974 * combine.c (can_change_dest_mode): Reject changes in
2975 REGMODE_NATURAL_SIZE.
2977 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2979 * configure.ac (--enable-libssp): New.
2980 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2981 trying to determine target-specific default. Adjust indentation.
2982 * configure: Regenerated.
2983 * doc/install.texi (Configuration): Expand --disable-libssp
2986 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2988 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2991 2017-10-31 Martin Jambor <mjambor@suse.cz>
2994 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2996 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2998 * auto-profile.c (autofdo_source_profile::read): Use
2999 UNKNOWN_LOCATION rather than 0.
3000 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3001 (warning_at): ...this overload.
3002 (warning_at_rich_loc_n): Rename to...
3003 (warning_n): ...this overload.
3004 (error_at_rich_loc): Rename to...
3005 (error_at): ...this overload.
3006 (pedwarn_at_rich_loc): Rename to...
3007 (pedwarn): ...this overload.
3008 (permerror_at_rich_loc): Rename to...
3009 (permerror): ...this overload.
3010 (inform_at_rich_loc): Rename to...
3011 (inform): ...this overload.
3012 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3013 (diagnostic_n_impl_richloc): Rename to...
3014 (diagnostic_n_impl): ...this rich_location *-based decl.
3015 (inform_at_rich_loc): Rename to...
3016 (inform): ...this, and add an assertion.
3017 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3018 (warning_at_rich_loc): Rename to...
3019 (warning_at): ...this, and add an assertion.
3020 (warning_at_rich_loc_n): Rename to...
3021 (warning_n): ...this, and add an assertion.
3022 (warning_n): Update location_t-based implementation for removal of
3023 location_t-based diagnostic_n_impl.
3024 (pedwarn_at_rich_loc): Rename to...
3025 (pedwarn): ...this, and add an assertion.
3026 (permerror_at_rich_loc): Rename to...
3027 (permerror): ...this, and add an assertion.
3028 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3029 (error_at_rich_loc): Rename to...
3030 (error_at): ...this, and add an assertion.
3031 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3032 (driver::do_spec_on_infiles): Likewise.
3033 * substring-locations.c (format_warning_va): Update for renaming
3034 of inform_at_rich_loc.
3036 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3038 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3039 _Float<N>X built-in functions so that the variant without the
3040 "__builtin_" prefix is only enabled for the GNU C and Objective C
3041 languages when they are in non-strict ANSI/ISO mode.
3042 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3043 * target.def (floatn_builtin_p): Add a target hook to control
3044 whether _Float<N> and _Float<N>X built-in functions without the
3045 "__builtin_" prefix are enabled, and return true for C and
3046 Objective C in the default hook. Include langhooks.h in
3048 * targhooks.h (default_floatn_builtin_p): Likewise.
3049 * targhooks.c (default_floatn_builtin_p): Likewise.
3050 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3051 floatn_builtin_p target hook.
3052 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3054 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3055 Eric Botcazou <ebotcazou@adacore.com>
3057 PR rtl-optimization/81803
3058 * lra-constraints.c (curr_insn_transform): Also reload the whole
3059 register for a strict subreg no wider than a word if this is for
3060 a WORD_REGISTER_OPERATIONS target.
3062 2017-10-31 Jason Merrill <jason@redhat.com>
3064 * gdbinit.in: Skip over inlines from timevar.h.
3066 2017-10-31 Martin Liska <mliska@suse.cz>
3068 * doc/gcov.texi: Document new option.
3069 * gcov.c (print_usage): Likewise print it.
3070 (process_args): Support the argument.
3071 (format_count): New function.
3072 (format_gcov): Use the function.
3074 2017-10-31 Martin Liska <mliska@suse.cz>
3076 * gcov.c (struct name_map): do not use typedef.
3077 Define operator== and operator<.
3078 (name_search): Remove.
3079 (name_sort): Remove.
3080 (main): Do not allocate names.
3081 (process_file): Add vertical space.
3082 (generate_results): Use std::find.
3083 (release_structures): Do not release memory.
3084 (find_source): Use std::find.
3086 2017-10-31 Martin Liska <mliska@suse.cz>
3088 * gcov.c (struct line_info): Remove it's typedef.
3089 (line_info::line_info): Add proper ctor.
3090 (line_info::has_block): Do not use a typedef.
3091 (struct source_info): Do not use typedef.
3092 (circuit): Likewise.
3093 (get_cycles_count): Likewise.
3094 (output_intermediate_file): Iterate via vector iterator.
3095 (add_line_counts): Use std::vector methods.
3096 (accumulate_line_counts): Likewise.
3097 (output_lines): Likewise.
3099 2017-10-31 Martin Liska <mliska@suse.cz>
3101 * gcov.c (struct source_info): Remove typedef.
3102 (source_info::source_info): Add proper ctor.
3103 (accumulate_line_counts): Use struct, not it's typedef.
3104 (output_gcov_file): Likewise.
3105 (output_lines): Likewise.
3106 (main): Do not allocate an array.
3107 (output_intermediate_file): Use size of vector container.
3108 (process_file): Resize the vector.
3109 (generate_results): Do not preallocate, use newly added vector
3111 (release_structures): Do not release sources.
3112 (find_source): Use vector methods.
3113 (add_line_counts): Do not use typedef.
3115 2017-10-31 Martin Liska <mliska@suse.cz>
3117 * doc/gcov.texi: Document that.
3118 * gcov.c (add_line_counts): Mark lines with a non-executed
3120 (output_line_beginning): Handle such lines.
3121 (output_lines): Pass new argument.
3122 (output_intermediate_file): Print it in intermediate format.
3124 2017-10-31 Martin Liska <mliska@suse.cz>
3126 * color-macros.h: New file.
3127 * diagnostic-color.c: Factor out color related to macros to
3129 * doc/gcov.texi: Document -k option.
3130 * gcov.c (INCLUDE_STRING): Include string.h.
3131 (print_usage): Add -k option.
3132 (process_args): Parse it.
3133 (pad_count_string): New function.
3134 (output_line_beginning): Likewise.
3135 (DEFAULT_LINE_START): New macro.
3136 (output_lines): Support color output.
3138 2017-10-31 Martin Liska <mliska@suse.cz>
3140 PR gcov-profile/82633
3141 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3142 their interaction with GCOV infrastructure.
3143 * configure.ac: Add -fkeep-{inline,static}-functions to
3145 * configure: Regenerate.
3147 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3150 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3152 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3155 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3156 into a register if it does not fit into an immediate offset field.
3158 2017-10-31 Olivier Hainque <hainque@adacore.com>
3160 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3162 2017-10-31 Julia Koval <julia.koval@intel.com>
3164 * config.gcc: Add gfniintrin.h.
3165 * config/i386/gfniintrin.h: New.
3166 * config/i386/i386-builtin-types.def
3167 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3168 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3169 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3170 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3171 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3172 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3173 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3174 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3175 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3176 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3177 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3178 * config/i386/immintrin.h: Include gfniintrin.h.
3179 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3181 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3183 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3185 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3187 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3188 (arm_ashldi3_1bit): Remove pattern.
3189 (ashrdi3): Remove shift by 1 expansion.
3190 (arm_ashrdi3_1bit): Remove pattern.
3191 (lshrdi3): Remove shift by 1 expansion.
3192 (arm_lshrdi3_1bit): Remove pattern.
3193 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3194 cost of ashldi3 by 1.
3195 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3196 (<shift>di3_neon): Likewise.
3198 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3200 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3201 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3202 and (*aarch64_simd_mov<VQ:mode>).
3203 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3204 pattern alternative.
3205 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3206 attributes to match pattern alternative.
3208 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3210 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3211 * config/rs6000/emmintrin.h: New file.
3212 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3214 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3216 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3217 * (movdi_vfp_cortexa8): Remove pattern.
3219 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3221 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3223 (Specific, alpha*-dec-osf5.1): Remove.
3224 (Specific, mips-sgi-irix5): Remove.
3225 (Specific, mips-sgi-irix6): Remove.
3227 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3230 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3231 arguments to clear_bit_region_be.
3233 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3235 * gimplify.c: Include memmodel.h.
3237 2017-10-30 Martin Jambor <mjambor@suse.cz>
3239 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3240 condition whether loop should be executed at all.
3242 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3245 gimple folding of vec_madd() intrinsics.
3246 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3247 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3248 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3250 2017-10-30 Richard Biener <rguenther@suse.de>
3252 PR tree-optimization/82762
3254 2017-10-23 Richard Biener <rguenther@suse.de>
3256 PR tree-optimization/82129
3258 2017-08-01 Richard Biener <rguenther@suse.de>
3260 PR tree-optimization/81181
3261 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3262 (compute_antic): ... end of iteration here.
3264 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3266 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3267 -std=iso9899:2017 and -std=gnu17.
3268 * doc/standards.texi (C Language): Document C17 support.
3269 * doc/cpp.texi (Overview): Mention -std=c17.
3270 (Standard Predefined Macros): Document C11 and C17 values of
3271 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3272 * doc/extend.texi (Inline): Do not list individual options for
3273 standards newer than C99.
3274 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3276 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3279 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3281 * asan.c (asan_finish_file): Align asan globals array by shadow
3284 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3287 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3288 (struct store_immediate_info): Add bitregion_start and bitregion_end
3290 (store_immediate_info::store_immediate_info): Add brs and bre
3291 arguments and initialize bitregion_{start,end} from those.
3292 (struct merged_store_group): Add bitregion_start, bitregion_end,
3293 align_base and mask fields. Drop unnecessary struct keyword from
3294 struct store_immediate_info. Add do_merge method.
3295 (clear_bit_region_be): Use memset instead of loop storing zeros.
3296 (merged_store_group::do_merge): New method.
3297 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3298 stores as long as the surrounding bitregions have no gaps.
3299 (merged_store_group::merge_overlapping): Use do_merge.
3300 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3301 is byte aligned, rather than requiring that start and width are
3302 byte aligned. Drop unnecessary struct keyword from
3303 struct store_immediate_info. Allocate and populate also mask array.
3304 Make start of the arrays relative to bitregion_start rather than
3305 start and size them according to bitregion_{end,start} difference.
3306 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3307 struct store_immediate_info.
3308 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3309 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3310 struct keyword from struct store_immediate_info.
3311 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3312 between stores as long as the surrounding bitregions have no gaps.
3314 (struct split_store): Add orig non-static data member.
3315 (split_store::split_store): Initialize orig to false.
3316 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3317 if there is exactly a single original stmt. Change stmts argument
3318 to pointer from reference, if NULL, don't push anything to it. Add
3319 first argument, use it to optimize skipping over orig stmts that
3320 are known to be before bitpos already. Simplify.
3321 (split_group): Return unsigned int count how many stores are or
3322 would be needed rather than a bool. Add allow_unaligned argument.
3323 Change split_stores argument from reference to pointer, if NULL,
3324 only do a dry run computing how many stores would be produced.
3325 Rewritten algorithm to use both alignment and misalign if
3326 !allow_unaligned and handle bitfield stores with gaps.
3327 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3328 from bitregion_start instead of start. Compute allow_unaligned
3329 here, if true, do 2 split_group dry runs to compute which one
3330 produces fewer stores and prefer aligned if equal. Punt if
3331 new count is bigger or equal than original before emitting any
3332 statements, rather than during that. Remove no longer needed
3333 new_ssa_names tracking. Replace num_stmts with
3334 split_stores.length (). Use 32-bit stack allocated entries
3335 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3336 if possible. Handle bitfields with gaps.
3337 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3338 Compute bitregion_{start,end} for the stores and construct
3339 store_immediate_info with that. Formatting fixes.
3341 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3344 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3345 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3347 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3349 * gimplify.c: Include tm_p.h.
3351 * common.opt (gcoff): Re-add as ignored option.
3352 (gcoff1, gcoff2, gcoff3): Likewise.
3354 * Makefile.in (OBJS): Delete sdbout.o.
3355 (GTFILES): Delete $(srcdir)/sdbout.c.
3356 * debug.h: Delete sdb_debug_hooks.
3357 * final.c: Delete sdbout.h include.
3358 (final_scan_insn): Delete SDB_DEBUG check.
3359 (rest_of_clean_state): Likewise.
3360 * output.h: Delete sdb_begin_function_line.
3363 * toplev.c: Delete sdbout.h include.
3364 (process_options): Delete SDB_DEBUG check.
3365 * tree-core.h (tree_type_common): Delete pointer field of
3367 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3368 TYPE_SYMTAB_POINTER.
3369 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3370 (TYPE_SYMTAB_IS_POINTER): Delete.
3371 (TYPE_SYMTAB_IS_DIE): Renumber.
3372 * xcoffout.c: Refer to former sdbout.c file.
3373 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3375 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3376 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3377 (Debugging Options): Delete -gcoff.
3378 (-gstabs): Delete SDB reference.
3380 (-gcoff@var{level}): Delete.
3381 * doc/passes.texi (Debugging information output): Delete SDB and
3382 sdbout.c references.
3383 * doc/tm.texi: Regenerate.
3384 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3385 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3387 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3388 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3389 (SDB_DEBUGGING_INFO): Delete.
3390 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3391 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3392 * target.def (output_source_filename): Delete COFF reference.
3394 * common.opt (gcoff): Delete.
3395 (gxcoff+): Update Negative chain.
3396 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3398 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3399 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3400 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3402 (expand_function_start): Change sdb reference to past tense.
3403 (expand_function_end): Change sdb reference to past tense.
3404 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3405 * opts.c (debug_type_names): Delete coff entry.
3406 (common_handle_option): Delete OPT_gcoff case.
3407 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3409 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3410 * config/cris/cris.h: Delete SDB reference in comment.
3411 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3412 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3413 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3414 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3416 (ix86_expand_prologue): Likewise.
3417 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3418 * config/ia64/ia64.h: Likewise.
3419 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3420 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3422 * config/mmix/mmix.h: Likewise.
3423 * config/nds32/nds32.c: Likewise.
3424 * config/stormy/storym16.h: Likewise.
3425 * config/visium/visium.h: Likewise.
3426 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3428 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3430 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3431 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3432 FRAME_POINTER_REGNUM point at high end of local var area.
3434 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3436 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3437 Move comment around. Do not reset best_edge for a copiable
3438 destination if the copy would cause a partition change.
3439 (better_edge_p): Remove redundant check.
3441 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3443 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3445 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3447 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3448 for math functions that have _Float<N> and _Float<N>X variants.
3449 (mathfn_built_in_2): Add support for math functions that have
3450 _Float<N> and _Float<N>X variants.
3451 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3452 (expand_builtin_mathfn_ternary): Add support for fma with
3453 _Float<N> and _Float<N>X variants.
3454 (expand_builtin): Likewise.
3455 (fold_builtin_3): Likewise.
3456 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3457 create math function _Float<N> and _Float<N>X variants as external
3459 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3460 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3461 the __builtin_ prefix and if not strict ansi, without the prefix.
3462 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3463 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3464 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3465 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3466 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3467 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3468 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3469 function signatures for fma _Float<N> and _Float<N>X variants.
3470 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3471 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3472 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3473 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3474 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3475 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3476 * gencfn-macros.c (print_case_cfn): Add support for math functions
3477 that have _Float<N> and _Float<N>X variants.
3478 (print_define_operator_list): Likewise.
3479 (fltfn_suffixes): Likewise.
3481 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3482 for math functions that have _Float<N> and _Float<N>X variants.
3483 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3484 and _Float<N>X variants.
3485 (COPYSIGN): Likewise.
3488 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3489 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3491 (integer_valued_read_call_p): Likewise.
3492 * fold-const-call.c (fold_const_call_ss): Likewise.
3493 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3494 _Float<N> and _Float<N>X variants.
3495 (fold_const_call_ssss): Add support for fma _Float<N> and
3496 _Float<N>X variants.
3497 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3498 support for copysign and fma _Float<N> and _Float<N>X variants.
3499 (backprop::process_builtin_call_use): Likewise.
3500 * tree-call-cdce.c (can_test_argument_range); Add support for
3501 sqrt _Float<N> and _Float<N>X variants.
3502 (edom_only_function): Likewise.
3503 (get_no_error_domain): Likewise.
3504 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3505 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3506 copysign _Float<N> and _Float<N>X variants.
3507 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3508 handled by machine independent code.
3509 (FMAF128): Likewise.
3510 * doc/cpp.texi (Common Predefined Macros): Document defining
3511 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3512 fma _Float<N> and _Float<N>X variants.
3514 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3517 * config/i386/i386-modes.def (CCFPU): Remove definition.
3518 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3519 (ix86_cc_modes_compatible): Ditto.
3520 (ix86_expand_carry_flag_compare): Ditto.
3521 (ix86_expand_int_movcc): Ditto.
3522 (ix86_expand_int_addcc): Ditto.
3523 (ix86_reverse_condition): Ditto.
3524 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3525 Return true/false for unordered/ordered fp comparisons.
3526 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3527 (ix86_prepare_fp_compare_args): Update for rename.
3528 (ix86_expand_fp_compare): Update for rename. Generate unordered
3529 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3530 (ix86_expand_sse_compare_and_jump): Ditto.
3531 * config/i386/predicates.md (fcmov_comparison_operator):
3532 Remove CCFPU mode handling.
3533 (ix86_comparison_operator): Ditto.
3534 (ix86_carry_flag_operator): Ditto.
3535 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3536 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3537 (*cmpu<mode>_cc_i387): Ditto.
3538 (FPCMP): Remove mode iterator.
3539 (unord): Remove mode attribute.
3540 (unord_subst): New define_subst transformation
3541 (unord): New define_subst attribute.
3543 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3544 (*cmpi<unord>xf_i387): Ditto.
3545 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3546 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3547 using unord_subst transformation.
3548 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3549 (round_saeonly): Also handle CCFP mode.
3550 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3551 Remove UNSPEC_SAHF unspec handling.
3553 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3555 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3557 2017-10-27 Jeff Law <law@redhat.com>
3559 * gimple-ssa-sprintf.c: Include domwalk.h.
3560 (class sprintf_dom_walker): New class, derived from dom_walker.
3561 (sprintf_dom_walker::before_dom_children): New function.
3562 (struct call_info): Moved into sprintf_dom_walker class
3563 (compute_formath_length, handle_gimple_call): Likewise.
3564 (sprintf_length::execute): Call the dominator walker rather
3565 than walking the statements.
3567 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3568 gimple statement locations.
3569 (check_array_bounds): Corresponding changes. Get the statement's
3570 location directly from wi->stmt.
3572 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3575 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3577 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3579 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3580 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3582 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3585 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3586 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3587 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3588 maybe_get_pool_constant.
3589 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3592 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3594 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3595 2.26 caveat. Update gas and gld versions.
3596 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3599 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3601 * cgraph.h (set_malloc_flag): Declare.
3602 * cgraph.c (set_malloc_flag_1): New function.
3603 (set_malloc_flag): Likewise.
3604 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3605 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3607 (read_ipa_call_summary): Add support for reading is_return_callee.
3608 (write_ipa_call_summary): Stream is_return_callee.
3609 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3610 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3611 ipa-prop.h, ipa-fnsummary.h.
3612 (pure_const_names): Change to static.
3613 (malloc_state_e): Define.
3614 (malloc_state_names): Define.
3615 (funct_state_d): Add field malloc_state.
3616 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3617 (check_retval_uses): New function.
3618 (malloc_candidate_p): Likewise.
3619 (analyze_function): Add support for malloc attribute.
3620 (pure_const_write_summary): Stream malloc_state.
3621 (pure_const_read_summary): Add support for reading malloc_state.
3622 (dump_malloc_lattice): New function.
3623 (propagate_malloc): New function.
3624 (warn_function_malloc): New function.
3625 (ipa_pure_const::execute): Call propagate_malloc and
3626 ipa_free_fn_summary.
3627 (pass_local_pure_const::execute): Add support for malloc attribute.
3628 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3629 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3631 2017-10-27 Martin Liska <mliska@suse.cz>
3633 PR gcov-profile/82457
3634 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3635 for fork-like functions to be properly instrumented.
3637 2017-10-27 Richard Biener <rguenther@suse.de>
3640 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3641 info when we redirected EH.
3643 2017-10-26 Michael Collison <michael.collison@arm.com>
3645 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3647 (<optab>_trunchf<GPI:mode>2: New pattern.
3648 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3649 * config/aarch64/iterators.md (wv): New mode attribute.
3650 (vf, VF): New mode attributes.
3651 (vgp, VGP): New mode attributes.
3652 (s): Update attribute with SImode and DImode prefixes.
3654 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3656 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3657 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3658 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3659 (nios2_option_overide): Initialize it. Don't allow R0-relative
3660 addressing with PIC.
3661 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3662 (nios2_symbolic_constant_p): Likewise.
3663 (nios2_legitimate_address_p): Likewise.
3664 (nios2_r0rel_section_name_p): New.
3665 (nios2_symbol_ref_in_r0rel_data_p): New.
3666 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3667 (r0rel_constant_p): New.
3668 (nios2_print_operand_address): Handle r0rel_constant_p.
3669 (nios2_cdx_narrow_form_p): Likewise.
3670 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3671 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3672 (Nios II Options): Document -mr0rel-sec.
3674 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3676 * config/nios2/nios2.c: Include xregex.h.
3677 (nios2_gprel_sec_regex): New.
3678 (nios2_option_overide): Initialize it. Don't allow GP-relative
3679 addressing with PIC.
3680 (nios2_small_section_name_p): Check for regex match.
3681 * config/nios2/nios2.opt (mgprel-sec=): New option.
3682 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3683 (Nios II Options): Document -mgprel-sec.
3685 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3687 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3689 2017-10-26 Tom de Vries <tom@codesourcery.com>
3691 PR tree-optimization/82707
3692 * gimple.c (gimple_copy): Fix unsharing of
3693 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3695 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3697 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3698 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3699 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3700 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3701 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3702 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3703 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3704 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3705 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3706 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3707 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3708 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3709 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3710 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3711 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3712 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3713 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3715 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3717 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3719 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3720 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3721 warning to rs6000.c. Remove the Undocumented flag, since it has
3723 (-mabi=ibmlongdouble): Likewise.
3724 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3725 already set, set the default format for long double.
3726 (rs6000_debug_reg_global): Print whether long double is IBM or
3728 (rs6000_option_override_internal): Rework setting long double
3729 format. Only warn if the user is changing the long double default
3730 and they did not use -Wno-psabi.
3731 * doc/invoke.texi (PowerPC options): Update the documentation for
3732 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3734 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3735 Alan Hayward <alan.hayward@arm.com>
3736 David Sherwood <david.sherwood@arm.com>
3738 * rtl.h (wider_subreg_mode): New function.
3739 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3740 rather than an unsigned int *.
3741 * ira-color.c (regno_max_ref_width): Replace with...
3742 (regno_max_ref_mode): ...this new variable.
3743 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3744 Use wider_subreg_mode.
3745 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3746 rather than an unsigned int *.
3747 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3748 (process_alt_operands): Likewise.
3749 (invariant_p): Likewise.
3750 * lra-spills.c (assign_mem_slot): Likewise.
3751 (add_pseudo_to_slot): Likewise.
3752 * lra.c (collect_non_operand_hard_regs): Likewise.
3753 (add_regs_to_insn_regno_info): Likewise.
3754 * reload1.c (regno_max_ref_width): Replace with...
3755 (regno_max_ref_mode): ...this new variable.
3756 (reload): Update accordingly. Update call to
3757 ira_sort_regnos_for_alter_reg.
3758 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3759 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3760 (scan_paradoxical_subregs): Likewise.
3762 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3764 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3765 (aarch64_frame): Add emit_frame_chain boolean.
3766 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3767 Move eh_return case to aarch64_layout_frame.
3768 (aarch64_layout_frame): Initialize emit_frame_chain.
3769 (aarch64_expand_prologue): Use emit_frame_chain.
3771 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3773 * config/aarch64/aarch64.c (aarch64_layout_frame):
3774 Ensure LR is always stored at the bottom of the callee-saves.
3775 Remove rarely used frame layout which saves callee-saves at top of
3776 frame, so the store of LR can be used as a valid probe in all cases.
3778 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3780 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3781 Improve unaligned TImode/TFmode base/offset split.
3783 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3784 Alan Hayward <alan.hayward@arm.com>
3785 David Sherwood <david.sherwood@arm.com>
3787 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3788 * combine.c (find_single_use_1): Likewise.
3789 (expand_field_assignment): Likewise.
3790 (move_deaths): Likewise.
3791 * lra-constraints.c (simplify_operand_subreg): Likewise.
3792 (curr_insn_transform): Likewise.
3793 * lra.c (collect_non_operand_hard_regs): Likewise.
3794 (add_regs_to_insn_regno_info): Likewise.
3795 * rtlanal.c (reg_referenced_p): Likewise.
3796 (covers_regno_no_parallel_p): Likewise.
3798 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3800 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3801 Don't print any bits outside the precision of the value.
3802 * wide-int.cc (test_printing): Add some new tests.
3804 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3806 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3807 supports -xbrace_comment option.
3808 * configure: Regenerate.
3809 * config.in: Regenerate.
3810 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3811 (ASM_CPU_SPEC): Use it.
3813 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3815 * target.def (static_rtx_alignment): New hook.
3816 * targhooks.h (default_static_rtx_alignment): Declare.
3817 * targhooks.c (default_static_rtx_alignment): New function.
3818 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3819 * doc/tm.texi: Regenerate.
3820 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3821 instead of targetm.constant_alignment. Remove call to
3823 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3824 (cris_preferred_mininum_alignment): New function, split out from...
3825 (cris_constant_alignment): ...here.
3826 (cris_static_rtx_alignment): New function.
3827 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3829 (ix86_constant_alignment): ...here.
3830 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3831 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3832 (mmix_static_rtx_alignment): New function.
3833 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3834 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3836 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3839 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3840 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3842 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3844 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3845 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3846 mark as a sign-extending load.
3847 (local_pic_load_u): Define.
3849 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3852 * fold-const.c (operand_equal_for_comparison_p): Also return true
3853 if ARG0 is a simple variant of ARG1 with narrower precision.
3854 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3856 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3858 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3860 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3861 scatter_static, scatter_per_elt.
3862 * x86-tune-costs.h: Add new cost entries.
3864 2017-10-25 Richard Biener <rguenther@suse.de>
3866 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3867 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3868 class pass_fre): Move to ...
3869 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3870 class pass_fre): ... here and adjust for statistics.
3872 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3875 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3876 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3877 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3878 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3881 2017-10-25 Richard Biener <rguenther@suse.de>
3883 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3884 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3885 eliminate_push_avail, eliminate_insert): Move inside...
3886 (class eliminate_dom_walker): ... this class in preparation
3888 (fini_eliminate): Remove by merging with ...
3889 (eliminate): ... this function. Adjust for class changes.
3890 (pass_pre::execute): Remove fini_eliminate call.
3891 (pass_fre::execute): Likewise.
3893 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3896 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3897 (VPERMI2, VPERMI2I): New mode iterators.
3898 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3899 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3901 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3902 mode iterator. Remove 3 old define_insn patterns.
3903 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3904 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3905 VPERMI2 mode iterator, remove the other two expanders.
3906 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3907 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3908 instructions, remove the other two patterns.
3909 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3910 mode iterator, remove the other two patterns.
3911 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3912 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3913 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3914 and adjust argument order accordingly.
3915 (ix86_expand_vec_perm): Adjust caller.
3916 (expand_vec_perm_1): Likewise.
3917 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3918 (expand_vec_perm_vpermt2_vpshub2): ... this.
3919 (ix86_expand_vec_perm_const_1): Adjust caller.
3920 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3923 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3924 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3925 (vec_shl_<mode>): Remove unused expander.
3926 (avx512bw_<shift_insn><mode>3): New define_insn.
3927 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3928 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3930 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3933 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3936 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3938 PR rtl-optimization/82396
3939 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3940 (autopref_multipass_init): Simplify initialization.
3941 (autopref_rank_data): Simplify sort order.
3942 * gcc/sched-int.h (autopref_multipass_data_): Remove
3943 multi_mem_insn_p, min_offset and max_offset.
3945 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3948 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3949 Check special value of flag_omit_frame_pointer.
3950 (aarch64_can_eliminate): Likewise.
3951 (aarch64_override_options_after_change_1): Simplify handling of
3952 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3954 2017-10-24 Richard Biener <rguenther@suse.de>
3956 PR tree-optimization/82697
3957 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3958 zero for conditional load and unconditional store.
3960 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3962 * doc/install.texi: Document bootstrap-cet.
3964 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3967 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3968 ENDBR instruction at function entrance if function is only
3971 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3974 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3975 patterns to better describe from which operation the CF is computed.
3976 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3977 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3978 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3979 is 0, use _0 suffixed expanders instead of emitting a comparison
3982 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3984 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3985 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3987 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3990 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3991 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3992 * loop-iv.c (iv_get_reaching_def): Likewise.
3993 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3994 variable is promoted and the partition contains undefined values.
3996 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3998 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4000 (nios2_address_cost): Define.
4001 (nios2_legitimize_address): Recognize (exp + constant) directly.
4002 (TARGET_ADDRESS_COST): Define.
4004 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4006 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4007 (nios2_symbolic_memory_operand_p): Declare.
4008 (nios2_split_large_constant): Declare.
4009 (nios2_split_symbolic_memory_operand): Declare.
4010 * config/nios2/nios2.c: Adjust includes.
4011 (nios2_symbolic_constant_allowed): New.
4012 (nios2_symbolic_constant_p): New.
4013 (nios2_plus_symbolic_constant_p): New.
4014 (nios2_valid_addr_expr_p): Recognize addresses involving
4016 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4017 (nios2_symbolic_memory_operand_p): New.
4018 (nios2_large_constant_p): New.
4019 (nios2_split_large_constant): New.
4020 (nios2_split_plus_large_constant): New.
4021 (nios2_split_symbolic_memory_operand): New.
4022 (nios2_legitimize_address): Code refactoring. Handle addresses
4023 involving symbolic constants.
4024 (nios2_emit_move_sequence): Likewise.
4025 (nios2_print_operand): Improve error output.
4026 (nios2_print_operand_address): Handle LO_SUM.
4027 (nios2_cdx_narrow_form_p): Likewise.
4028 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4029 operands involving symbolic constants.
4030 (movhi_internal, movsi_internal): Likewise.
4031 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4032 (extendhisi2, extendqi<mode>2): Likewise.
4034 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4036 * tree-pass.h (PROP_rtl_split_insns): Define.
4037 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4039 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4041 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4043 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4046 * target.def (const_not_ok_for_debug_p): Default to
4047 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4048 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4049 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4050 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4051 which targetm.const_not_ok_for_debug_p returned true.
4052 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4054 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4056 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4057 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4058 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4060 (ix86_const_not_ok_for_debug_p): New function.
4061 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4062 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4064 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4067 * system.h: Conditionally include "unique-ptr.h" if
4068 INCLUDE_UNIQUE_PTR is defined.
4069 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4070 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4072 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4074 * config/rl78/rl78.md: New define_expand "subdi3".
4076 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4079 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4080 DF_REF_INSN if DF_REF_INSN_INFO is false.
4082 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4084 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4085 xmm_move instead of sse_move.
4086 (sse_store_index): New function.
4087 (ix86_register_move_cost): Be more sensible about mismatch stall;
4088 model AVX moves correctly; make difference between sse->integer and
4090 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4091 moves; make difference between SSE and AVX.
4092 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4093 and zmm_move. Increase size of sse load and store tables;
4094 add unaligned load and store tables; add ssemmx_to_integer.
4095 * x86-tune-costs.h: Update all entries according to real
4096 move latencies from Agner Fog's manual and chip documentation.
4098 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4101 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4102 * config/i386/constraints.md (Wf): New constraint.
4103 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4104 (cmp<dwi>_doubleword): Removed.
4105 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4106 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4107 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4108 expand with cmp<dwi>_doubleword. For LTU and GEU use
4109 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4111 * common.opt (gcolumn-info): Enable by default.
4112 * doc/invoke.texi (gcolumn-info): Document new default.
4114 2017-10-23 Richard Biener <rguenther@suse.de>
4116 PR tree-optimization/82672
4117 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4118 Fold the stmt if we propagated into it.
4120 2017-10-23 Richard Biener <rguenther@suse.de>
4122 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4123 (bitmap_remove_expr_from_set): ... this. All callers call this
4124 for non-constant values.
4125 (bitmap_set_subtract): Rename to...
4126 (bitmap_set_subtract_expressions): ... this. Adjust and
4128 (bitmap_set_contains_value): Remove superfluous check.
4129 (bitmap_set_replace_value): Inline into single caller ...
4130 (bitmap_value_replace_in_set): ... here and simplify.
4131 (dependent_clean): Merge into ...
4132 (clean): ... this using an overload. Adjust.
4133 (prune_clobbered_mems): Adjust.
4134 (compute_antic_aux): Likewise.
4135 (compute_partial_antic_aux): Likewise.
4137 2017-10-23 Richard Biener <rguenther@suse.de>
4139 PR tree-optimization/82129
4141 2017-08-01 Richard Biener <rguenther@suse.de>
4143 PR tree-optimization/81181
4144 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4145 (compute_antic): ... end of iteration here.
4147 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4149 * target.def (starting_frame_offset): New hook.
4150 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4151 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4152 * doc/tm.texi.in: Regenerate.
4153 * hooks.h (hook_hwi_void_0): Declare.
4154 * hooks.c (hook_hwi_void_0): New function.
4155 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4156 STARTING_FRAME_OFFSET.
4157 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4158 * reload1.c (reload): Likewise.
4159 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4160 instead of STARTING_FRAME_OFFSET.
4161 * function.c (try_fit_stack_local): Likewise.
4162 (assign_stack_local_1): Likewise
4163 (instantiate_virtual_regs): Likewise.
4164 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4165 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4166 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4167 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4168 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4169 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4170 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4171 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4172 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4173 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4174 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4175 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4176 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4177 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4178 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4179 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4180 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4181 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4182 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4183 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4184 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4185 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4186 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4187 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4188 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4189 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4190 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4191 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4192 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4193 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4194 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4195 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4196 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4197 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4198 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4199 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4200 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4201 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4202 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4203 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4204 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4205 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4206 return a HOST_WIDE_INT.
4207 (avr_builtin_setjmp_frame_value): Use it instead of
4208 STARTING_FRAME_OFFSET.
4209 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4210 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4211 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4213 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4214 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4215 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4216 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4217 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4218 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4219 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4220 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4221 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4222 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4223 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4224 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4226 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4227 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4228 * config/mips/mips.c (mips_compute_frame_info): Refer to
4229 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4230 (mips_starting_frame_offset): New function.
4231 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4232 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4233 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4234 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4235 and return a HOST_WIDE_INT.
4236 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4237 (mmix_initial_elimination_offset): Refer to
4238 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4239 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4240 * config/pa/pa.c (pa_starting_frame_offset): New function.
4241 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4242 (pa_expand_prologue): Likewise.
4243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4244 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4245 !FRAME_GROWS_DOWNWARD handling to...
4246 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4247 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4248 !FRAME_GROWS_DOWNWARD handling to...
4249 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4250 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4251 !FRAME_GROWS_DOWNWARD handling to...
4252 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4253 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4255 (rs6000_starting_frame_offset): New function.
4256 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4257 !FRAME_GROWS_DOWNWARD handling to...
4258 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4259 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4260 !FRAME_GROWS_DOWNWARD handling to...
4261 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4262 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4263 !FRAME_GROWS_DOWNWARD handling to...
4264 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4265 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4266 (rs6000_starting_frame_offset): New function.
4267 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4268 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4269 * config/vax/vax.c (vax_starting_frame_offset): New function.
4270 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4271 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4272 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4273 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4274 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4275 * system.h (STARTING_FRAME_OFFSET): Poison.
4277 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4279 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4280 SCALAR_TYPE_MODE instead of TYPE_MODE.
4282 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4283 Alan Hayward <alan.hayward@arm.com>
4284 David Sherwood <david.sherwood@arm.com>
4286 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4288 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4289 Alan Hayward <alan.hayward@arm.com>
4290 David Sherwood <david.sherwood@arm.com>
4292 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4294 2017-10-23 Richard Biener <rguenther@suse.de>
4296 PR tree-optimization/82129
4297 * tree-ssa-pre.c (bitmap_set_and): Remove.
4298 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4299 canonicalizing expressions in the set to those with lowest
4300 ID rather than taking that from the first edge.
4302 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4304 * combine.c (rtx_equal_for_field_assignment_p): Use
4305 byte_lowpart_offset.
4307 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4308 Alan Hayward <alan.hayward@arm.com>
4309 David Sherwood <david.sherwood@arm.com>
4311 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4312 to a SUBREG_PROMOTED_VAR.
4314 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4315 Alan Hayward <alan.hayward@arm.com>
4316 David Sherwood <david.sherwood@arm.com>
4318 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4319 (expand_debug_source_expr): Likewise.
4320 * combine.c (combine_simplify_rtx): Likewise.
4321 * cse.c (fold_rtx): Likewise.
4322 * optabs.c (expand_float): Likewise.
4323 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4324 (simplify_binary_operation_1): Likewise.
4326 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4327 Alan Hayward <alan.hayward@arm.com>
4328 David Sherwood <david.sherwood@arm.com>
4330 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4331 (record_promoted_value): Likewise.
4332 * expr.c (expand_expr_real_2): Likewise.
4333 * ree.c (update_reg_equal_equiv_notes): Likewise.
4334 (combine_set_extension): Likewise.
4335 * rtlanal.c (low_bitmask_len): Likewise.
4336 * simplify-rtx.c (neg_const_int): Likewise.
4337 (simplify_binary_operation_1): Likewise.
4339 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4340 Alan Hayward <alan.hayward@arm.com>
4341 David Sherwood <david.sherwood@arm.com>
4343 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4344 * regcprop.c (maybe_mode_change): Likewise.
4345 * reload1.c (alter_reg): Likewise.
4347 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4349 * inchash.h (inchash::hash::add_wide_int): New function.
4350 * lto-streamer-out.c (hash_tree): Use it.
4352 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4354 * inchash.h (inchash::hash::add_wide_int): Rename to...
4355 (inchash::hash::add_hwi): ...this.
4356 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4357 (polymorphic_call_target_hasher::hash): Likewise.
4358 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4359 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4360 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4361 * lto-streamer-out.c (hash_tree): Likewise.
4362 * optc-save-gen.awk: Likewise.
4363 * tree.c (add_expr): Likewise.
4365 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4368 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4369 for ordered inequality comparisons even with TARGET_IEEE_FP.
4371 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4374 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4375 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4376 Expand with cmp<dwi>_doubleword.
4378 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4380 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4381 List CET intrinsics.
4382 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4383 specific to -fcf-protection option.
4385 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4387 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4388 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4389 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4390 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4391 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4392 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4393 (extra_objs): Add cet.o for Linux/x86 targets.
4394 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4395 * config/i386/cet.c: New file.
4396 * config/i386/cetintrin.h: Likewise.
4397 * config/i386/t-cet: Likewise.
4398 * config/i386/cpuid.h (bit_SHSTK): New.
4399 (bit_IBT): Likewise.
4400 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4401 pass IBT and SHSTK bits.
4402 * config/i386/i386-builtin-types.def
4403 (VOID_FTYPE_UNSIGNED_PVOID): New.
4404 (VOID_FTYPE_UINT64_PVOID): Likewise.
4405 * config/i386/i386-builtin.def: Add CET intrinsics.
4406 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4407 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4408 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4409 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4411 * config/i386/i386.c (rest_of_insert_endbranch): New.
4412 (pass_data_insert_endbranch): Likewise.
4413 (pass_insert_endbranch): Likewise.
4414 (make_pass_insert_endbranch): Likewise.
4415 (ix86_notrack_prefixed_insn_p): Likewise.
4416 (ix86_target_string): Add -mibt, -mshstk flags.
4417 (ix86_option_override_internal): Add flag_cf_protection
4419 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4420 (ix86_print_operand): Add 'notrack' prefix output.
4421 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4422 (ix86_expand_builtin): Expand CET intrinsics.
4423 (x86_output_mi_thunk): Add 'endbranch' instruction.
4424 * config/i386/i386.h (TARGET_IBT): New.
4425 (TARGET_IBT_P): Likewise.
4426 (TARGET_SHSTK): Likewise.
4427 (TARGET_SHSTK_P): Likewise.
4428 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4429 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4430 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4431 (builtin_setjmp_setup): New pattern.
4432 (builtin_longjmp): Likewise.
4433 (rdssp<mode>): Likewise.
4434 (incssp<mode>): Likewise.
4435 (saveprevssp): Likewise.
4436 (rstorssp): Likewise.
4437 (wrss<mode>): Likewise.
4438 (wruss<mode>): Likewise.
4439 (setssbsy): Likewise.
4440 (clrssbsy): Likewise.
4441 (nop_endbr): Likewise.
4442 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4444 * config/i386/immintrin.h: Include <cetintrin.h>.
4445 * config/i386/linux-common.h
4446 (file_end_indicate_exec_stack_and_cet): New prototype.
4447 (TARGET_ASM_FILE_END): New.
4449 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4451 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4452 latencies instead of having separate table; make difference between
4453 integer and float costs.
4454 * i386.h (processor_costs): Remove scalar_stmt_cost,
4455 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4456 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4458 * x86-tune-costs.h: Remove entries which has been removed in
4459 procesor_costs from all tables; make cond_taken_branch_cost
4460 and cond_not_taken_branch_cost COST_N_INSNS based.
4462 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4464 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4466 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4468 * config/i386/i386.md (isa): Remove fma_avx512f.
4469 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4470 <avx512>_fmadd_<mode>_mask3<round_name>,
4471 <avx512>_fmsub_<mode>_mask<round_name>,
4472 <avx512>_fmsub_<mode>_mask3<round_name>,
4473 <avx512>_fnmadd_<mode>_mask<round_name>,
4474 <avx512>_fnmadd_<mode>_mask3<round_name>,
4475 <avx512>_fnmsub_<mode>_mask<round_name>,
4476 <avx512>_fnmsub_<mode>_mask3<round_name>,
4477 <avx512>_fmaddsub_<mode>_mask<round_name>,
4478 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4479 <avx512>_fmsubadd_<mode>_mask<round_name>,
4480 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4481 (*vec_widen_umult_even_v16si<mask_name>,
4482 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4483 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4485 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4487 * extend.texi: Add 'nocf_check' documentation.
4488 * gimple.texi: Add second parameter to
4489 gimple_build_call_from_tree.
4490 * invoke.texi: Add -fcf-protection documentation.
4491 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4493 2017-10-20 Richard Biener <rguenther@suse.de>
4495 PR tree-optimization/82473
4496 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4497 the largest input type.
4499 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4501 * c-attribs.c (handle_nocf_check_attribute): New function.
4502 (c_common_attribute_table): Add 'nocf_check' handling.
4503 * gimple-parser.c: Add second argument NULL to
4504 gimple_build_call_from_tree.
4505 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4506 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4508 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4509 * common.opt: Add fcf-protection flag.
4510 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4511 * flag-types.h: Add enum cf_protection_level.
4512 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4513 Add 'nocf_check' attribute propagation to gimple call.
4514 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4515 (gimple_build_call_from_tree): Update prototype.
4516 (gimple_call_nocf_check_p): New function.
4517 (gimple_call_set_nocf_check): Likewise.
4518 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4519 * ipa-icf.c: Add nocf_check attribute in statement hash.
4520 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4521 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4522 * toplev.c (process_options): Add flag_cf_protection handling.
4524 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4526 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4528 2017-10-20 Richard Biener <rguenther@suse.de>
4530 PR tree-optimization/82603
4531 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4532 remove false predicated stores.
4534 2017-10-20 Richard Biener <rguenther@suse.de>
4536 * graphite-isl-ast-to-gimple.c
4537 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4538 Remove return value and simplify, dump copied stmt after lhs
4540 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4541 Reduce dump verbosity.
4542 (gsi_insert_earliest): Likewise.
4543 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4544 * graphite.c (print_global_statistics): Adjust dumping.
4545 (print_graphite_scop_statistics): Likewise.
4546 (print_graphite_statistics): Do not dump loops here.
4547 (graphite_transform_loops): But here.
4549 2017-10-20 Nicolas Roche <roche@adacore.com>
4551 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4552 * configure: Regenerate.
4554 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4557 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4558 functions when optimizing replace GIMPLE_RETURN stmts with
4559 calls to __builtin_unreachable ().
4562 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4563 for -fsanitize=thread link of executables.
4564 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4565 link of executables.
4568 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4570 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4571 define_insns for logical vector shifts to use VI248_AVX512BW
4572 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4573 condition, useless isa and prefix attributes. Change the first
4574 2 of these define_insns to ...
4575 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4576 define_insn for avx512vl.
4577 (<shift_insn><mode>3): ... and this, new define_insn without
4578 masking for non-avx512vl.
4581 * config/i386/sse.md (*andnot<mode>3,
4582 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4583 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4584 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4585 not applied use empty suffix even for TARGET_AVX512VL.
4586 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4587 is applied, supply evex,evex or evex,evex,evex instead of just
4590 2017-10-20 Julia Koval <julia.koval@intel.com>
4592 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4593 (OPTION_MASK_ISA_GFNI_UNSET): New.
4594 (ix86_handle_option): Handle OPT_mgfni.
4595 * config/i386/cpuid.h (bit_GFNI): New.
4596 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4597 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4598 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4599 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4600 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4601 * config/i386/i386.opt: Add mgfni.
4603 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4605 * config/msp430/msp430.c (msp430_option_override): Disable
4606 -fdelete-null-pointer-checks.
4607 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4609 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4611 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4612 of x87 and SSE instructions.
4614 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4616 * asan.c (create_cond_insert_point): Do not update edge count.
4617 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4618 (afdo_propagate_circuit): Likewise.
4619 (afdo_calculate_branch_prob): Likewise.
4620 (afdo_annotate_cfg): Likewise.
4621 * basic-block.h (struct edge_def): Remove count.
4622 (edge_def::count): New accessor.
4623 * bb-reorder.c (rotate_loop): Update.
4624 (find_traces_1_round): Update.
4625 (connect_traces): Update.
4626 (sanitize_hot_paths): Update.
4627 * cfg.c (unchecked_make_edge): Update.
4628 (make_single_succ_edge): Update.
4629 (check_bb_profile): Update.
4630 (dump_edge_info): Update.
4631 (update_bb_profile_for_threading): Update.
4632 (scale_bbs_frequencies_int): Update.
4633 (scale_bbs_frequencies_gcov_type): Update.
4634 (scale_bbs_frequencies_profile_count): Update.
4635 (scale_bbs_frequencies): Update.
4636 * cfganal.c (connect_infinite_loops_to_exit): Update.
4637 * cfgbuild.c (compute_outgoing_frequencies): Update.
4638 (find_many_sub_basic_blocks): Update.
4639 * cfgcleanup.c (try_forward_edges): Update.
4640 (try_crossjump_to_edge): Update
4641 * cfgexpand.c (expand_gimple_cond): Update
4642 (expand_gimple_tailcall): Update
4643 (construct_exit_block): Update
4644 * cfghooks.c (verify_flow_info): Update
4645 (redirect_edge_succ_nodup): Update
4646 (split_edge): Update
4647 (make_forwarder_block): Update
4648 (duplicate_block): Update
4649 (account_profile_record): Update
4650 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4651 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4652 * cfgloopmanip.c (scale_loop_profile): Update.
4654 (lv_adjust_loop_entry_edge): Update.
4655 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4656 (force_nonfallthru_and_redirect): Update.
4657 (purge_dead_edges): Update.
4658 (rtl_flow_call_edges_add): Update.
4659 * cgraphunit.c (init_lowered_empty_function): Update.
4660 (cgraph_node::expand_thunk): Update.
4661 * gimple-pretty-print.c (dump_probability): Update.
4662 (dump_edge_probability): Update.
4663 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4664 * haifa-sched.c (sched_create_recovery_edges): Update.
4665 * hsa-gen.c (convert_switch_statements): Update.
4666 * ifcvt.c (dead_or_predicable): Update.
4667 * ipa-inline-transform.c (inline_transform): Update.
4668 * ipa-split.c (split_function): Update.
4669 * ipa-utils.c (ipa_merge_profiles): Update.
4670 * loop-doloop.c (add_test): Update.
4671 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4672 * lto-streamer-in.c (input_cfg): Update.
4673 (input_function): Update.
4674 * lto-streamer-out.c (output_cfg): Update.
4675 * modulo-sched.c (sms_schedule): Update.
4676 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4677 * predict.c (maybe_hot_edge_p): Update.
4678 (unlikely_executed_edge_p): Update.
4679 (probably_never_executed_edge_p): Update.
4680 (dump_prediction): Update.
4681 (drop_profile): Update.
4682 (propagate_unlikely_bbs_forward): Update.
4683 (determine_unlikely_bbs): Update.
4684 (force_edge_cold): Update.
4685 * profile.c (compute_branch_probabilities): Update.
4686 * reg-stack.c (better_edge): Update.
4687 * shrink-wrap.c (handle_simple_exit): Update.
4688 * tracer.c (better_p): Update.
4689 * trans-mem.c (expand_transaction): Update.
4690 (split_bb_make_tm_edge): Update.
4691 * tree-call-cdce.c: Update.
4692 * tree-cfg.c (gimple_find_sub_bbs): Update.
4693 (gimple_split_edge): Update.
4694 (gimple_duplicate_sese_region): Update.
4695 (gimple_duplicate_sese_tail): Update.
4696 (gimple_flow_call_edges_add): Update.
4697 (insert_cond_bb): Update.
4698 (execute_fixup_cfg): Update.
4699 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4700 * tree-complex.c (expand_complex_div_wide): Update.
4701 * tree-eh.c (lower_resx): Update.
4702 (unsplit_eh): Update.
4703 (cleanup_empty_eh_move_lp): Update.
4704 * tree-inline.c (copy_edges_for_bb): Update.
4705 (freqs_to_counts): Update.
4706 (copy_cfg_body): Update.
4707 * tree-ssa-dce.c (remove_dead_stmt): Update.
4708 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4709 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4710 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4711 (unloop_loops): Update.
4712 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4713 * tree-ssa-loop-split.c (connect_loops): Update.
4714 (split_loop): Update.
4715 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4716 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4717 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4718 * tree-ssa-reassoc.c (branch_fixup): Update.
4719 * tree-ssa-tail-merge.c (replace_block_by): Update.
4720 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4721 (compute_path_counts): Update.
4722 (update_profile): Update.
4723 (recompute_probabilities): Update.
4724 (update_joiner_offpath_counts): Update.
4725 (estimated_freqs_path): Update.
4726 (freqs_to_counts_path): Update.
4727 (clear_counts_path): Update.
4728 (ssa_fix_duplicate_block_edges): Update.
4729 (duplicate_thread_path): Update.
4730 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4731 (case_bit_test_cmp): Update.
4732 (collect_switch_conv_info): Update.
4733 (gen_inbound_check): Update.
4734 (do_jump_if_equal): Update.
4735 (emit_cmp_and_jump_insns): Update.
4736 * tree-tailcall.c (decrease_profile): Update.
4737 (eliminate_tail_call): Update.
4738 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4739 (vect_do_peeling): Update.
4740 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4741 * ubsan.c (ubsan_expand_null_ifn): Update.
4742 (ubsan_expand_ptr_ifn): Update.
4743 * value-prof.c (gimple_divmod_fixed_value): Update.
4744 (gimple_mod_pow2): Update.
4745 (gimple_mod_subtract): Update.
4746 (gimple_ic): Update.
4747 (gimple_stringop_fixed_value): Update.
4749 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4752 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4754 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4756 PR rtl-optimization/82395
4757 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4758 based on non_spilled_static_chain_regno_p.
4760 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4762 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4763 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4764 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4766 2017-10-19 Martin Sebor <msebor@redhat.com>
4768 PR tree-optimization/82596
4769 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4771 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4773 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4774 (get_mem_refs_of_builtin_call): Likewise.
4775 * builtins.c (expand_builtin_apply): Adjust call to
4776 allocate_dynamic_stack_space.
4777 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4778 the third argument to allocate_dynamic_stack_space, otherwise -1.
4779 (expand_builtin): Deal with all alloca variants.
4780 (is_inexpensive_builtin): Likewise.
4781 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4782 * calls.c (special_function_p): Deal with all alloca variants.
4783 (initialize_argument_information): Adjust call to
4784 allocate_dynamic_stack_space.
4785 (expand_call): Likewise.
4786 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4787 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4788 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4789 use it for the stack usage computation.
4790 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4791 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4792 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4793 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4794 (in_loop_p): Remove first argument and useless check.
4795 (pass_walloca::execute): Remove useless test and adjust call to above.
4796 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4797 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4798 (gimplify_call_expr): Deal with all alloca variants.
4799 * hsa-gen.c (gen_hsa_alloca): Likewise.
4800 (gen_hsa_insns_for_call): Likewise.
4801 * ipa-pure-const.c (special_builtin_state): Likewise.
4802 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4803 * tree-object-size.c (alloc_object_size): Likewise.
4804 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4805 (call_may_clobber_ref_p_1): Likewise.
4806 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4807 (ccp_fold_stmt): Likewise.
4808 (optimize_stack_restore): Likewise.
4809 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4810 (mark_all_reaching_defs_necessary_1): Likewise.
4811 (propagate_necessity): Likewise.
4812 (eliminate_unnecessary_stmts): Likewise.
4813 * tree.c (build_common_builtin_nodes): Build
4814 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4815 (build_alloca_call_expr): New function.
4816 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4817 (CASE_BUILT_IN_ALLOCA): Likewise.
4818 (build_alloca_call_expr): Declare.
4819 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4821 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4824 * dwarf2out.c (new_die_raw): New static inline function.
4825 (new_die): Use it to create the DIE.
4826 (add_AT_external_die_ref): Likewise.
4827 (clone_die): Likewise.
4828 (clone_as_declaration): Likewise.
4829 (dwarf2out_vms_debug_main_pointer): Likewise.
4830 (base_type_die): Likewise. Remove early return for corner cases.
4831 Do not call add_pubtype on the DIE here.
4832 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4833 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4834 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4835 native order exists for base types, attach the DIE manually and call
4836 add_pubtype on it. Do not equate a reverse order DIE to the type.
4838 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4840 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4841 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4843 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4845 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4847 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4848 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4849 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4850 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4851 * opts.c (sanitizer_opts): Add builtin.
4852 * ubsan.c (instrument_builtin): New function.
4853 (pass_ubsan::execute): Call it.
4854 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4855 * doc/invoke.texi: Document -fsanitize=builtin.
4857 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4858 builtins, store max (log2 (align), 0) into uchar field instead of
4859 align into uptr field.
4860 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4861 store uchar 0 field instead of uptr 0 field.
4862 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4863 instead of passing one address of struct with 2 locations pass
4864 two addresses of structs with 1 location each.
4865 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4866 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4867 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4868 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4869 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4870 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4871 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4872 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4874 2017-10-19 Martin Liska <mliska@suse.cz>
4877 * file-find.c (remove_prefix): Remove.
4878 * file-find.h (remove_prefix): Likewise.
4879 * gcc-ar.c: Remove smartness of lookup.
4881 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4883 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4884 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4885 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4887 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4890 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4891 (setcc + and to xor + setcc): New peephole2.
4893 2017-10-19 Tom de Vries <tom@codesourcery.com>
4895 * doc/sourcebuild.texi (Test Directives, Variants of
4896 dg-require-support): Add dg-require-stack-size.
4898 2017-10-19 Martin Liska <mliska@suse.cz>
4901 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4902 that have a large alignment.
4903 (gimplify_target_expr): Likewise.
4905 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4907 PR rtl-optimization/82602
4908 * ira.c (rtx_moveable_p): Return false for volatile asm.
4910 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4913 * config/i386/i386-modes.def (CCGZ): New CC mode.
4914 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4915 * config/i386/predicates.md (ix86_comparison_operator):
4917 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4918 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4919 with double-word subtraction.
4920 (put_condition_code): Handle CCGZmode.
4922 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4924 * wide-int.cc (debug (const wide_int &)): New.
4925 (debug (const wide_int *)): New.
4926 (debug (const widest_int &)): New.
4927 (debug (const widest_int *)): New.
4929 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4932 * lra-constraints.c (curr_insn_transform): Use non-input operand
4933 instead of output one for matched reload.
4935 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4937 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4938 (tree-ssa-loop-ivopts.h): New header file.
4939 (struct builtin_info): New fields.
4940 (classify_builtin_1): Compute and record base and offset parts for
4941 memset builtin partition by calling strip_offset.
4942 (offset_cmp, fuse_memset_builtins): New functions.
4943 (finalize_partitions): Fuse adjacent memset partitions by calling
4945 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4946 Expose the interface.
4947 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4949 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4951 PR tree-optimization/82574
4952 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4953 that data reference must be executed exactly once per iteration
4954 against the outermost loop in nest.
4955 (classify_partition): Update call to above function.
4957 2017-10-18 Richard Biener <rguenther@suse.de>
4959 PR tree-optimization/82591
4960 * graphite.c (graphite_transform_loops): Move code gen message
4962 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4963 Here. Handle scop_to_isl_ast failing.
4964 (scop_to_isl_ast): Limit the number of ISL operations.
4966 2017-10-18 Richard Biener <rguenther@suse.de>
4968 * graphite-isl-ast-to-gimple.c
4969 (translate_isl_ast_to_gimple::set_rename): Simplify.
4970 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4971 (graphite_copy_stmts_from_block): ... here.
4972 (copy_bb_and_scalar_dependences): Simplify.
4973 (add_parameters_to_ivs_params): Canonicalize.
4974 (generate_entry_out_of_ssa_copies): Simplify.
4975 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4976 by passing in ISL dimension.
4977 (parameter_index_in_region_1): Rename to ...
4978 (parameter_index_in_region): ... this.
4979 (extract_affine): Adjust assert, pass down parameter index.
4980 (add_param_constraints): Use range-info when available.
4981 (build_scop_context): Adjust.
4982 * sese.c (new_sese_info): Adjust.
4983 (free_sese_info): Likewise.
4984 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4985 Remove unused typedefs.
4986 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4988 2017-10-18 Martin Liska <mliska@suse.cz>
4990 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4992 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4994 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4995 (s390_sched_init): Do not reset s390_sched_state if we entered the
4996 current basic block via a fallthru edge and all others are unlikely.
4998 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5000 * config/s390/s390.c (NUM_SIDES): New variable.
5001 (LONGRUNNING_THRESHOLD): New variable.
5002 (LATENCY_FACTOR): New variable.
5003 (s390_sched_score): Decrease score for long-running instructions on
5005 (s390_sched_variable_issue): Perform bookkeeping for long-running
5008 2017-10-18 Richard Biener <rguenther@suse.de>
5010 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5011 Simplify with removal of the parameter rename map.
5012 (set_rename): Likewise.
5013 (should_copy_to_new_region): Likewise.
5014 (graphite_copy_stmts_from_block): Likewise.
5015 (copy_bb_and_scalar_dependences): Remove initialization of
5016 unused copied_bb_map.
5018 (copy_internal_parameters): Likewise.
5019 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5020 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5021 Use INTEGRAL_TYPE_P.
5022 (parameter_index_in_region_1): Rename to ...
5023 (assign_parameter_index_in_region): ... this. Assert we have
5024 a parameter we handle.
5025 (scan_tree_for_params): Adjust.
5026 * sese.h (parameter_rename_map_t): Remove.
5027 (struct sese_info_t): Remove unused parameter_rename_map and
5028 copied_bb_map members.
5029 * sese.c (new_sese_info): Adjust.
5030 (free_sese_info): Likewise.
5032 2017-10-18 Martin Liska <mliska@suse.cz>
5035 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5036 on an abnormal edge.
5038 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5040 * doc/invoke.texi (ffunction-sections and fdata-sections):
5043 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5045 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5046 the use statement can throw internally.
5048 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5050 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5051 any RTX present on the RHS of a SET.
5052 * compare-elim.c (try_eliminate_compare): Restore comment.
5054 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5056 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5057 may be also called on tcc_type nodes.
5058 * langhooks.c (lhd_tree_size): Likewise.
5060 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5062 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5063 format_warning_at_substring.
5064 (maybe_warn): Convert source_range * param to a location_t. Pass
5065 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5066 (format_directive): Remove code to extract source_ranges and
5067 source_range * in favor of just a location_t.
5068 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5070 * substring-locations.c (format_warning_va): Convert
5071 source_range * param to a location_t.
5072 (format_warning_at_substring): Likewise.
5073 * substring-locations.h (format_warning_va): Likewise.
5074 (format_warning_at_substring): Likewise.
5076 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5078 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5080 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5081 and scatter/gather ops.
5083 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5084 vec_gather_load and vec_scatter_store.
5085 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5086 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5087 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5088 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5089 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5090 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5092 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5094 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5095 Detect FTST instruction and handle its register pops. Only pop
5096 second operand if can_pop_second_op is true.
5097 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5098 set can_pop_second_op to false in the compare_for_stack_reg call.
5100 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5101 output_fp_compare for stack register operands.
5102 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5103 instructions here. Do not emit stack register pops here. Assert
5104 that FCOMPP pops next to top stack register. Rewrite function.
5106 2017-10-17 Nathan Sidwell <nathan@acm.org>
5109 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5110 use DECL_ASSEMBLER_NAME_RAW.
5113 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5116 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5117 Wilco Dijkstra <wilco.dijkstra@arm.com>
5119 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5120 converstion to Pmode from the buf_addr.
5122 2017-10-17 Richard Biener <rguenther@suse.de>
5124 * graphite-dependences.c (scop_get_reads_and_writes): Change
5125 output parameters to references.
5127 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5129 PR 71026/tree-optimization
5130 * fold-const.c (distribute_real_division): Removed.
5131 (fold_binary_loc): Remove calls to distribute_real_divison.
5133 2017-10-17 Richard Biener <rguenther@suse.de>
5135 * graphite-scop-detection.c
5136 (scop_detection::stmt_has_simple_data_refs_p): Always use
5137 the full nest as region.
5138 (try_generate_gimple_bb): Likewise.
5139 * sese.c (scalar_evolution_in_region): Simplify now that
5140 SCEV can handle instantiation in regions.
5141 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5142 in the non-loop part of a function if requested.
5144 2017-10-17 Richard Biener <rguenther@suse.de>
5146 PR tree-optimization/82563
5147 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5149 (graphite_regenerate_ast_isl): Call it.
5150 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5152 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5154 PR tree-optimization/82549
5155 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5156 Formatting fixes. Instead of calling make_bit_field_ref with negative
5159 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5161 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5162 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5163 _mm_maskz_reduce_ss): New.
5164 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5165 __builtin_ia32_reducess_mask): Ditto..
5166 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5167 * config/i386/sse.md (reduces<mode>): Renamed to ...
5168 (reduces<mode><mask_scalar_name>): ... this.
5169 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5171 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5172 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5174 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5176 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5177 * selftest-run-tests.c (selftest::run_tests): Call
5178 selftest::unique_ptr_tests_cc_tests.
5179 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5180 * unique-ptr-tests.cc: New file.
5182 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5185 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5187 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5188 (make_hard_regno_born, make_hard_regno_dead): Update
5189 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5191 2017-10-16 Jeff Law <law@redhat.com>
5193 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5195 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5197 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5199 2017-10-16 Olivier Hainque <hainque@adacore.com>
5201 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5202 with_cpu if we were configured for an e500v2 target cpu name.
5204 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5206 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5207 * doc/invoke.texi: Document +nodsp as a valid extension for
5210 2017-10-16 Martin Liska <mliska@suse.cz>
5212 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5213 (test_set_range): Likewise.
5214 (test_range_functions): Rename to ...
5215 (test_bit_in_range): ... this.
5216 (sbitmap_c_tests): Add new test.
5218 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5220 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5222 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5223 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5225 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5227 * config/aarch64/aarch64-builtins.c
5228 (aarch64_types_quadopu_lane_qualifiers): New.
5229 (TYPES_QUADOPU_LANE): New.
5230 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5231 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5232 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5233 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5234 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5235 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5236 (Vdottype, DOTPROD): New.
5237 (sur): Add SDOT and UDOT.
5239 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5241 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5242 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5243 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5245 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5246 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5247 Enable TARGET_DOTPROD.
5248 (cortex-a75.cortex-a55): Likewise.
5249 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5251 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5253 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5254 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5256 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5258 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5259 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5260 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5261 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5262 * config/arm/types.md (neon_dot, neon_dot_q): New.
5263 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5265 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5267 * config/arm/arm.h (TARGET_DOTPROD): New.
5268 * config/arm/arm.c (arm_arch_dotprod): New.
5269 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5270 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5271 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5272 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5273 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5274 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5275 * doc/invoke.texi (armv8.2-a): Document dotprod
5277 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5279 * i386.c (ix86_vec_cost): New function.
5280 (ix86_rtx_costs): Handle vector operations better.
5281 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5282 * x86-tune-costs.h: Add new costs to all tables.
5284 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5286 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5288 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5289 divsd, sqrtss and sqrtsd
5290 * x86-tune-costs.h: Add new entries to all costs.
5291 (znver1_cost): Fix to match real instruction latencies.
5293 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5294 Michael Collison <michael.collison@arm.com>
5296 * compare-elim.c: Include emit-rtl.h.
5297 (can_merge_compare_into_arith): New function.
5298 (try_validate_parallel): Likewise.
5299 (try_merge_compare): Likewise.
5300 (try_eliminate_compare): Call the above when no previous clobber
5302 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5305 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5309 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5310 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5314 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5315 to be any operand_equal_p operands. For & (B - 1) require
5316 B to be power of 2. Recognize
5317 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5319 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5322 * optabs.c (expand_memory_blockage): Fix call of
5323 targetm.have_memory_blockage.
5325 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5328 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5329 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5330 objects to extra_objs instead of overwriting it.
5332 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5334 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5335 Use any_fp_register_operand as operand[3] predicate. Simplify
5336 equality test for operands[2] and operands[4] memory location.
5337 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5338 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5339 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5340 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5341 any_fp_register_operand as operand[1] predicate. Simplify
5342 equality test for operands[0] and operands[3] memory location.
5343 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5344 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5345 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5347 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5349 * target-insns.def: Add memory_blockage.
5350 * optabs.c (expand_memory_blockage): New function.
5351 (expand_asm_memory_barrier): Rename ...
5352 (expand_asm_memory_blockage): ... to this.
5353 (expand_mem_thread_fence): Call expand_memory_blockage
5354 instead of expand_asm_memory_barrier.
5355 (expand_mem_singnal_fence): Ditto.
5356 (expand_atomic_load): Ditto.
5357 (expand_atomic_store): Ditto.
5358 * doc/md.texi (Standard Pattern Names For Generation):
5359 Document memory_blockage instruction pattern.
5361 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5363 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5364 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5365 * config/rl78/rl78.md: New define_expand "adddi3".
5367 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5369 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5372 2017-10-13 Jeff Law <law@redhat.com>
5374 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5376 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5379 * internal-fn.c (expand_mul_overflow): If both operands have
5380 the same highpart of -1 or 0 and the topmost bit of lowpart
5381 is different, overflow is if res <= 0 rather than res < 0.
5383 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5385 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5386 TARGET_P9_VECTOR code for unaligned_load case.
5388 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5390 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5392 2017-10-13 Nathan Sidwell <nathan@acm.org>
5394 * tree-core.h (tree_contains_struct): Make bool.
5395 * tree.c (tree_contains_struct): Likewise.
5396 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5397 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5398 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5399 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5400 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5402 2017-10-13 Richard Biener <rguenther@suse.de>
5404 * graphite-isl-ast-to-gimple.c
5405 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5406 parameters and dominance check.
5407 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5408 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5409 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5410 Do not update SSA form here or do intermediate IL verification.
5411 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5412 (graphite_initialize): Remove check on the number of loops in
5413 the function and inline into graphite_transform_loops.
5414 (graphite_finalize): Inline into graphite_transform_loops.
5415 (graphite_transform_loops): Perform SSA update and IL verification
5417 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5419 2017-10-13 Richard Biener <rguenther@suse.de>
5421 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5422 graphite_expression_type_precision): Avoid global constructor
5424 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5425 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5426 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5427 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5428 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5429 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5431 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5434 * config/i386/i386.h (ix86_red_zone_size): New.
5435 * config/i386/i386.md (push peephole2s): Replace
5436 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5438 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5439 Alan Hayward <alan.hayward@arm.com>
5440 David Sherwood <david.sherwood@arm.com>
5442 * combine.c (can_change_dest_mode): Reject changes in
5443 REGMODE_NATURAL_SIZE.
5445 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5450 (expand_debug_source_expr): Likewise.
5451 * combine.c (combine_simplify_rtx): Likewise.
5452 * cse.c (fold_rtx): Likewise.
5453 * fwprop.c (canonicalize_address): Likewise.
5454 * targhooks.c (default_shift_truncation_mask): Likewise.
5456 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5457 Alan Hayward <alan.hayward@arm.com>
5458 David Sherwood <david.sherwood@arm.com>
5460 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5461 (widened_mode): Likewise.
5462 (expand_unop): Likewise.
5463 * ree.c (transform_ifelse): Likewise.
5464 (merge_def_and_ext): Likewise.
5465 (combine_reaching_defs): Likewise.
5466 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5468 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5469 Alan Hayward <alan.hayward@arm.com>
5470 David Sherwood <david.sherwood@arm.com>
5472 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5473 * combine.c (gen_lowpart_for_combine): Likewise.
5474 * dwarf2out.c (rtl_for_decl_location): Likewise.
5475 * final.c (alter_subreg): Likewise.
5476 * rtlhooks.c (gen_lowpart_general): Likewise.
5477 (gen_lowpart_if_possible): Likewise.
5479 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5480 Alan Hayward <alan.hayward@arm.com>
5481 David Sherwood <david.sherwood@arm.com>
5483 * calls.c (expand_call): Use subreg_lowpart_offset.
5484 * cse.c (cse_insn): Likewise.
5485 * regcprop.c (copy_value): Likewise.
5486 (copyprop_hardreg_forward_1): Likewise.
5488 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5491 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5492 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5493 =Q constraints to +Q and into insn condition add check
5494 that operands[0] and operands[1] are equal.
5495 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5496 =Q constraints to +Q and into insn condition add check
5497 that operands[0] is equal to either operands[1] or operands[2].
5500 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5501 instead of handling MINUS_EXPR twice (once for each argument),
5502 canonicalize operand order and handle just once, use rtype where
5503 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5506 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5507 any values of __C while still being pattern recognizable as a simple
5510 2017-10-13 Richard Biener <rguenther@suse.de>
5512 PR tree-optimization/82451
5514 2017-10-02 Richard Biener <rguenther@suse.de>
5516 PR tree-optimization/82355
5517 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5518 a mapping for the enclosing loop but avoid generating one for
5520 (copy_bb_and_scalar_dependences): Remove premature codegen
5521 error on PHIs in blocks duplicated into multiple places.
5522 * graphite-scop-detection.c
5523 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5524 in the region use it as loop and nest to analyze the DR in.
5525 (try_generate_gimple_bb): Likewise.
5526 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5527 (add_loop_constraints): For blocks in a loop not in the region
5528 create a dimension with a single iteration.
5529 * sese.h (gbb_loop_at_index): Remove assert.
5531 * cfgloop.c (loop_preheader_edge): For the loop tree root
5532 return the single successor of the entry block.
5533 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5534 Reset the SCEV hashtable and niters.
5535 * graphite-scop-detection.c
5536 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5537 assert that we only have POLYNOMIAL_CHREC that vary in loops
5538 contained in the region.
5539 (scop_detection::graphite_can_represent_expr): Adjust.
5540 (scop_detection::stmt_has_simple_data_refs_p): For loops
5541 not in the region set loop to NULL. The nest is now the
5542 entry edge to the region.
5543 (try_generate_gimple_bb): Likewise.
5544 * sese.c (scalar_evolution_in_region): Adjust for
5545 instantiate_scev change.
5546 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5547 Make nest parameter the edge into the region.
5548 (create_data_ref): Likewise.
5549 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5550 entry edge into a region and adjust instantiate_scev calls.
5551 (create_data_ref): Likewise.
5552 (graphite_find_data_references_in_stmt): Likewise.
5553 (find_data_references_in_stmt): Pass the loop preheader edge
5554 from the nest argument.
5555 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5556 parameter the edge into the region.
5557 (instantiate_parameters): Use the loop preheader edge as entry.
5558 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5560 (get_instantiated_value_entry): Make instantiate_below parameter
5561 the edge into the region.
5562 (instantiate_scev_name): Likewise. Adjust dominance checks,
5563 when we cannot use loop-based instantiation instantiate by
5564 walking use-def chains.
5565 (instantiate_scev_poly): Adjust.
5566 (instantiate_scev_binary): Likewise.
5567 (instantiate_scev_convert): Likewise.
5568 (instantiate_scev_not): Likewise.
5569 (instantiate_array_ref): Remove.
5570 (instantiate_scev_3): Likewise.
5571 (instantiate_scev_2): Likewise.
5572 (instantiate_scev_1): Likewise.
5573 (instantiate_scev_r): Do not blindly handle N-operand trees.
5574 Do not instantiate array-refs. Handle all constants and invariants.
5575 (instantiate_scev): Make instantiate_below parameter
5576 the edge into the region.
5577 (resolve_mixers): Use the loop preheader edge for the region
5578 parameter to instantiate_scev_r.
5579 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5581 2017-10-13 Richard Biener <rguenther@suse.de>
5583 PR tree-optimization/82525
5584 * graphite-isl-ast-to-gimple.c
5585 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5587 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5588 Fail code generation when we cannot represent the isl integer.
5589 (binary_op_to_tree): Elide modulo operations that are no-ops
5590 in the type we code generate. Remove now superfluous code
5593 2017-10-13 Richard Biener <rguenther@suse.de>
5595 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5596 (scop_detection::harmful_loop_in_region): Remove premature
5597 IV type restriction.
5598 (scop_detection::graphite_can_represent_scev): We can handle
5599 pointer IVs just fine.
5601 2017-10-13 Alan Modra <amodra@gmail.com>
5603 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5604 "Clobbers and Scratch Registers". Add paragraph on
5605 alternative to clobbers for scratch registers and OpenBLAS
5608 2017-10-13 Alan Modra <amodra@gmail.com>
5610 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5611 example of a memory input for a string of known length. Move
5612 commentary out of table. Add a number of new examples
5613 covering array memory inputs.
5615 2017-10-12 Martin Liska <mliska@suse.cz>
5617 PR tree-optimization/82493
5618 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5619 (test_range_functions): New function.
5620 (sbitmap_c_tests): Likewise.
5621 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5622 * selftest.h (sbitmap_c_tests): New function.
5624 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5626 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5628 * config/rs6000/amo.h: Fix spacing issue.
5630 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5633 * config/i386/i386.md (*ashl<mode>3_mask_1,
5634 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5635 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5638 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5640 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5641 (profile_probability): Set max_probability
5642 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5644 * profile-count.c (profile_probability::differs_from_p): Do not
5645 rely on max_probaiblity == 10000
5647 2017-10-12 Jeff Law <law@redhat.com>
5649 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5652 2017-10-12 Martin Sebor <msebor@redhat.com>
5656 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5657 (handle_alias_pairs): Call it.
5658 * common.opt (-Wattribute-alias): New option.
5659 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5660 * doc/invoke.texi (-Wattribute-alias): Document.
5662 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5665 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5667 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5669 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5670 (make_hard_regno_born, make_hard_regno_dead): Update
5671 bb_killed_pseudos and bb_gen_pseudos.
5673 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5675 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5677 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5679 * config/alpha/alpha.c (alpha_split_conditional_move):
5680 Use std::swap instead of manually swapping.
5681 (alpha_stdarg_optimize_hook): Ditto.
5682 (alpha_canonicalize_comparison): Ditto.
5684 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5686 * tree-loop-distribution.c (struct builtin_info): New struct.
5687 (struct partition): Refactor fields into struct builtin_info.
5688 (partition_free): Free struct builtin_info.
5689 (build_size_arg_loc, build_addr_arg_loc): Delete.
5690 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5691 information from struct builtin_info.
5692 (find_single_drs): New function refactored from classify_partition.
5693 Also moved builtin validity checks to this function.
5694 (compute_access_range, alloc_builtin): New functions.
5695 (classify_builtin_st, classify_builtin_ldst): New functions.
5696 (classify_partition): Refactor code into functions find_single_drs,
5697 classify_builtin_st and classify_builtin_ldst.
5698 (distribute_loop): Don't do runtime alias check when distributing
5700 (find_seed_stmts_for_distribution): New function.
5701 (pass_loop_distribution::execute): Refactor code finding seed
5702 stmts into above function. Support distribution for the innermost
5703 two-level loop nest. Adjust dump information.
5705 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5707 * tree-loop-distribution.c: Adjust the general comment.
5708 (NUM_PARTITION_THRESHOLD): New macro.
5709 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5710 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5711 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5712 in call to build_partition_graph.
5713 (finalize_partitions): New parameter. Make loop distribution more
5714 conservative by fusing more partitions.
5715 (distribute_loop): Don't do runtime alias check in case of loop nest
5717 (find_seed_stmts_for_distribution): New function.
5718 (prepare_perfect_loop_nest): New function.
5719 (pass_loop_distribution::execute): Refactor code finding seed stmts
5720 and loop nest into above functions. Support loop nest distribution.
5721 Adjust dump information accordingly.
5723 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5725 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5726 and set PTYPE_SEQUENTIAL for merged partition.
5728 2017-10-12 Richard Biener <rguenther@suse.de>
5730 PR tree-optimization/69728
5732 2017-09-19 Richard Biener <rguenther@suse.de>
5734 PR tree-optimization/69728
5735 * graphite-sese-to-poly.c (schedule_error): New global.
5736 (add_loop_schedule): Handle empty domain by failing the
5738 (build_original_schedule): Handle schedule_error.
5740 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5741 domain by returning an unchanged schedule.
5743 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5745 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5746 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5748 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5750 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5753 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5756 * expr.c (store_field): Don't optimize away bitsize == 0 store
5757 from CALL_EXPR with addressable return type.
5759 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5761 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5762 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5763 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5764 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5765 TARGET_ISEL instead of TARGET_ISEL<sel>.
5767 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5769 * config/rs6000/rs6000.c
5770 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5772 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5774 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5776 (reg_or_cint_operand): Fix comment.
5777 (reg_or_zero_operand): New predicate.
5778 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5779 * config/rs6000/rs6000.c (output_isel): Delete.
5780 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5781 instead of reg_or_cint_operand. Output instruction directly (not via
5783 (isel_unsigned_<mode>): Ditto.
5784 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5785 gpc_reg_operand. Add an instruction alternative for this. Output
5786 instruction directly.
5787 (*isel_reversed_unsigned_<mode>): Ditto.
5789 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5791 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5792 (TARGET_CANONICALIZE_COMPARISON): Define.
5794 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5797 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5798 Check whether the dest is REG before adding REG_EQUIV note.
5800 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5803 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5805 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5806 (make_hard_regno_born, make_hard_regno_dead): Update
5807 bb_killed_pseudos and bb_gen_pseudos.
5809 2017-10-11 Nathan Sidwell <nathan@acm.org>
5811 * incpath.h (enum incpath_kind): Name enum, prefix values.
5812 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5813 * incpath.c (heads, tails): Use INC_MAX.
5814 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5815 (merge_include_chains, split_quote_chain,
5816 register_include_chains): Update incpath_kind names.
5817 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5818 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5820 (add_framework_path, darwin_register_objc_includes): Likewise.
5821 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5823 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5825 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5826 Do not use float_operator operator predicate.
5827 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5828 * config/i386/predicates.md (float_operator): Remove predicate.
5830 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5832 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5833 (*jccxf_i387): Ditto.
5834 (*jcc<mode>_i387): Ditto.
5835 (*jccu<mode>_i387): Ditto.
5836 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5837 (*jcc_*_i387 splitters): Remove.
5838 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5839 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5840 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5843 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5845 * profile-count.h (slow_safe_scale_64bit): New function.
5846 (safe_scale_64bit): New inline.
5847 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5848 * profile-count.c: Include wide-int.h
5849 (slow_safe_scale_64bit): New.
5851 2017-10-11 Nathan Sidwell <nathan@acm.org>
5853 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5854 HAS_DECL_ASSEMBLER_NAME_P.
5855 * gimple-expr.c (gimple_decl_printable_name: Check
5856 HAS_DECL_ASSEMBLER_NAME_P too.
5857 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5858 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5859 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5860 * passes.c (rest_of_decl_compilation): Check
5861 HAS_DECL_ASSEMBLER_NAME_P too.
5862 * recog.c (verify_changes): Likewise.
5863 * tree-pretty-print.c (dump_decl_name): Likewise.
5864 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5866 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5867 (SET_DECL_ASSEMBLER_NAME): Use it.
5868 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5869 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5870 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5872 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5874 * config.gcc (i386, x86_64): Add extra objects.
5875 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5876 (ix86_min_insn_size): Declare.
5877 (ix86_issue_rate): Declare.
5878 (ix86_adjust_cost): Declare.
5879 (ia32_multipass_dfa_lookahead): Declare.
5880 (ix86_macro_fusion_p): Declare.
5881 (ix86_macro_fusion_pair_p): Declare.
5882 (ix86_bd_has_dispatch): Declare.
5883 (ix86_bd_do_dispatch): Declare.
5884 (ix86_core2i7_init_hooks): Declare.
5885 (ix86_atom_sched_reorder): Declare.
5886 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5887 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5888 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5889 (rip_relative_addr_p): Rename to ...
5890 (ix86_rip_relative_addr_p): ... this one; export.
5891 (memory_address_length): Update.
5892 (ix86_issue_rate): Move to x86-tune-sched.c.
5893 (ix86_flags_dependent): Move to x86-tune-sched.c.
5894 (ix86_agi_dependent): Move to x86-tune-sched.c.
5895 (exact_dependency_1): Move to x86-tune-sched.c.
5896 (exact_store_load_dependency): Move to x86-tune-sched.c.
5897 (ix86_adjust_cost): Move to x86-tune-sched.c.
5898 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5899 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5900 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5901 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5902 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5903 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5904 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5905 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5906 (min_insn_size): Rename to ...
5907 (ix86_min_insn_size): ... this one; export.
5908 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5909 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5910 (core2i7_first_cycle_multipass_backtrack): Move to
5911 x86-tune-sched-core.c.
5912 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5913 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5914 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5915 (ix86_avoid_jump_mispredicts): Update.
5916 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5917 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5918 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5919 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5920 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5921 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5922 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5923 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5924 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5925 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5926 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5927 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5928 (BIG): Move to ix86-tune-sched-bd.c.
5929 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5930 (enum insn_path): Move to ix86-tune-sched-bd.c.
5931 (get_mem_group): Move to ix86-tune-sched-bd.c.
5932 (is_cmp): Move to ix86-tune-sched-bd.c.
5933 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5934 (is_branch): Move to ix86-tune-sched-bd.c.
5935 (is_prefetch): Move to ix86-tune-sched-bd.c.
5936 (init_window): Move to ix86-tune-sched-bd.c.
5937 (allocate_window): Move to ix86-tune-sched-bd.c.
5938 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5939 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5940 (process_end_window): Move to ix86-tune-sched-bd.c.
5941 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5942 (find_constant): Move to ix86-tune-sched-bd.c.
5943 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5944 (has_immediate): Move to ix86-tune-sched-bd.c.
5945 (get_insn_path): Move to ix86-tune-sched-bd.c.
5946 (get_insn_group): Move to ix86-tune-sched-bd.c.
5947 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5948 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5949 (add_insn_window): Move to ix86-tune-sched-bd.c.
5950 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5951 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5952 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5953 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5954 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5955 (do_dispatch): Move to ix86-tune-sched-bd.c.
5956 (has_dispatch): Move to ix86-tune-sched-bd.c.
5957 * config/i386/t-i386: Add new object files.
5958 * config/i386/x86-tune-costs.h: New file.
5959 * config/i386/x86-tune-sched-atom.c: New file.
5960 * config/i386/x86-tune-sched-bd.c: New file.
5961 * config/i386/x86-tune-sched-core.c: New file.
5962 * config/i386/x86-tune-sched.c: New file.
5964 2017-10-11 Liu Hao <lh_mouse@126.com>
5966 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5967 the generic version below instead.
5968 (should_colorize): Recognize Windows consoles as terminals
5970 * pretty-print.c [__MINGW32__] (write_all): New function.
5971 [__MINGW32__] (find_esc_head): Likewise.
5972 [__MINGW32__] (find_esc_terminator): Likewise.
5973 [__MINGW32__] (eat_esc_sequence): Likewise.
5974 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5976 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5979 2017-10-11 Richard Biener <rguenther@suse.de>
5981 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5982 Properly call analyze_scalar_evolution with the loop of the stmt.
5984 2017-10-11 Richard Biener <rguenther@suse.de>
5986 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5987 * tree-core.h (tree_base): Add chrec_var union member.
5988 * tree.h (CHREC_VAR): Remove.
5989 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5990 * tree-chrec.h (build_polynomial_chrec): Adjust.
5991 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5992 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5994 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5996 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5997 * match.pd: ... here.
5998 ((T) X == (T) Y): Relax condition.
6000 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6002 PR tree-optimization/82472
6003 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6005 (break_alias_scc_partitions): Update postorder number.
6007 2017-10-11 Martin Liska <mliska@suse.cz>
6010 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6012 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6014 2017-10-11 Martin Liska <mliska@suse.cz>
6016 * print-rtl.c (print_insn): Move declaration of idbuf
6017 to same scope as name.
6019 2017-10-11 Martin Liska <mliska@suse.cz>
6024 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6027 2017-10-11 Martin Liska <mliska@suse.cz>
6030 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6033 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6035 * config.gcc (default_gnu_indirect_function): Default to yes for
6036 arm*-*-linux* with glibc.
6038 2017-10-11 Richard Biener <rguenther@suse.de>
6040 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6041 default-defs and types we do not want to analyze.
6042 (interpret_loop_phi): Replace unreachable code with an assert.
6043 (compute_scalar_evolution_in_loop): Remove and inline ...
6044 (analyze_scalar_evolution_1): ... here, replacing condition with
6045 what makes the intent clearer. Remove handling of cases
6046 get_scalar_evolution now handles.
6048 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6050 PR rtl-optimization/81434
6051 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6052 comment for main loop. In sched_group_found if, also add checks for
6053 pass and min_cost_group.
6055 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6057 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6058 (rs6000_insn_cost): New function.
6059 * config/rs6000/rs6000.md (cost): New attribute.
6061 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6062 H.J. Lu <hongjiu.lu@intel.com>
6066 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6067 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6068 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6069 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6070 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6071 OPTION_MASK_ISA_MMX and some other bit set require both
6072 mmx and the ISAs without the mmx bit.
6073 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6074 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6075 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6076 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6077 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6078 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6079 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6080 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6081 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6082 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6083 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6084 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6085 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6086 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6087 Add OPTION_MASK_ISA_MMX.
6089 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6091 * config.gcc (armv7*-*-freebsd*): New target.
6092 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6094 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6096 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6097 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6100 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6102 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6103 a template parameter.
6104 (WIDE_INT_REF_FOR): Update accordingly.
6105 * tree.h (wi::int_traits <const_tree>): Delete.
6106 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6107 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6108 (wi::tree_to_wide_ref): New typedef.
6109 (wi::to_wide): New function.
6110 * calls.c (get_size_range): Use wi::to_wide when operating on
6112 * cgraph.c (cgraph_node::create_thunk): Likewise.
6113 * config/i386/i386.c (ix86_data_alignment): Likewise.
6114 (ix86_local_alignment): Likewise.
6115 * dbxout.c (stabstr_O): Likewise.
6116 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6117 * expr.c (const_vector_from_tree): Likewise.
6118 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6119 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6120 (fold_negate_expr_1, int_const_binop_1, const_binop)
6121 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6122 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6123 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6124 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6125 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6126 (fold_not_const, round_up_loc): Likewise.
6127 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6128 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6129 (alloca_call_type): Likewise.
6130 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6131 * godump.c (go_output_typedef): Likewise.
6132 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6133 * internal-fn.c (get_min_precision): Likewise.
6134 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6135 * ipa-polymorphic-call.c
6136 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6137 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6138 (ipa_modify_call_arguments): Likewise.
6139 * match.pd: Likewise.
6140 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6141 * print-tree.c (print_node_brief, print_node): Likewise.
6142 * stmt.c (expand_case): Likewise.
6143 * stor-layout.c (layout_type): Likewise.
6144 * tree-affine.c (tree_to_aff_combination): Likewise.
6145 * tree-cfg.c (group_case_labels_stmt): Likewise.
6146 * tree-data-ref.c (dr_analyze_indices): Likewise.
6147 (prune_runtime_alias_test_list): Likewise.
6148 * tree-dump.c (dequeue_and_dump): Likewise.
6149 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6150 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6151 * tree-pretty-print.c (dump_generic_node): Likewise.
6152 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6153 (simple_iv_with_niters): Likewise.
6154 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6155 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6156 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6157 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6158 (refine_value_range_using_guard, number_of_iterations_ne_max)
6159 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6160 (get_cst_init_from_scev, record_nonwrapping_iv)
6161 (scev_var_range_cant_overflow): Likewise.
6162 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6163 * tree-ssa-pre.c (compute_avail): Likewise.
6164 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6165 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6166 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6167 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6168 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6169 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6170 (dump_case_nodes, try_switch_expansion): Likewise.
6171 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6172 (vect_do_peeling): Likewise.
6173 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6174 * tree-vect-stmts.c (vectorizable_load): Likewise.
6175 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6176 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6177 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6178 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6179 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6180 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6181 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6182 (evrp_dom_walker::before_dom_children): Likewise.
6183 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6184 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6185 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6186 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6187 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6188 (get_range_pos_neg): Likewise.
6189 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6190 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6191 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6192 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6193 * config/avr/avr.c (avr_fold_builtin): Likewise.
6194 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6195 * config/msp430/msp430.c (msp430_attr): Likewise.
6196 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6197 * config/powerpcspe/powerpcspe-c.c
6198 (altivec_resolve_overloaded_builtin): Likewise.
6199 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6200 (rs6000_expand_ternop_builtin): Likewise.
6201 * config/rs6000/rs6000-c.c
6202 (altivec_resolve_overloaded_builtin): Likewise.
6203 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6204 (rs6000_expand_ternop_builtin): Likewise.
6205 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6207 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6209 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6210 when copying loop nest with only one inner loop.
6212 2017-10-10 Richard Biener <rguenther@suse.de>
6214 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6215 blocks if SCEV is active.
6216 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6218 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6219 (scev_initialize): Assert we are not yet initialized.
6221 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6223 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6224 inner loop's exit stmt by making it always exit the loop, otherwise
6225 we would generate an infinite empty loop.
6227 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6229 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6230 renaming variables in new preheader if it's deleted.
6232 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6234 * tree-loop-distribution.c (struct partition): Remove unused field
6235 loops of the structure.
6236 (partition_alloc, partition_free): Ditto.
6237 (build_rdg_partition_for_vertex): Ditto.
6239 2017-10-09 Jeff Law <law@redhat.com>
6241 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6242 return type to match prototype and documentation.
6244 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6246 * config/rs6000/rs6000.c (processor_costs): Move to ...
6247 * config/rs6000/rs6000.h: ... here.
6248 (rs6000_cost): Declare.
6250 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6252 * except.c (setjmp_fn): New global variable.
6253 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6254 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6255 if DONT_USE_BUILTIN_SETJMP is defined.
6257 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6259 * target.def (insn_cost): New hook.
6260 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6261 * doc/tm.texi: Regenerate.
6262 * rtlanal.c (insn_cost): Use the new hook.
6264 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6266 * combine.c (combine_validate_cost): Compute the new insn_cost,
6267 not just pattern_cost.
6268 (try_combine): Adjust comment.
6270 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6272 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6274 * combine.c (uid_insn_cost): Adjust comment.
6275 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6277 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6278 * dse.c (find_shift_sequence): Ditto.
6279 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6280 (bb_valid_for_noce_process_p): Use pattern_cost.
6281 * rtl.h (insn_rtx_cost): Delete.
6282 (pattern_cost): New prototype.
6283 (insn_cost): New prototype.
6284 * rtlanal.c (insn_rtx_cost): Rename to...
6285 (pattern_cost): ... this.
6288 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6290 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6291 (*jcc<mode>_0_r_i387): Ditto.
6292 (*jccxf_r_i387): Ditto.
6293 (*jcc<mode>_r_i387): Ditto.
6294 (*jccu<mode>_r_i387): Ditto.
6295 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6296 (*jcc): Rename from *jcc_1.
6298 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6300 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6301 deferred rescans after the lvx/stvx recombination pre-pass.
6303 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6305 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6306 memory operation instruction support.
6307 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6308 (rs6000-ibm-aix[789]*): Likewise.
6309 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6310 Document new functions.
6312 2017-10-09 Richard Biener <rguenther@suse.de>
6314 PR tree-optimization/82397
6315 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6316 equality only for semantically equal trees.
6318 2017-10-09 Richard Biener <rguenther@suse.de>
6320 PR tree-optimization/82449
6321 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6322 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6323 allow constant addresses.
6324 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6327 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6329 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6332 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6335 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6338 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6341 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6343 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6346 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6347 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6349 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6351 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6352 (WI_BINARY_PREDICATE_RESULT): Likewise.
6353 (wi::binary_traits::operator_result): New type.
6354 (wi::binary_traits::predicate_result): Likewise.
6355 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6356 (generic_wide_int::operator==, generic_wide_int::operator!=)
6357 (generic_wide_int::operator&, generic_wide_int::and_not)
6358 (generic_wide_int::operator|, generic_wide_int::or_not)
6359 (generic_wide_int::operator^, generic_wide_int::operator+
6360 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6362 (operator~, unary operator-, operator==, operator!=, operator&)
6363 (operator|, operator^, operator+, binary operator-, operator*): New
6365 * expr.c (get_inner_reference): Use wi::bit_and_not.
6366 * fold-const.c (fold_binary_loc): Likewise.
6367 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6368 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6369 (bit_value_binop): Likewise.
6370 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6371 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6372 (extract_range_from_binary_expr_1): Likewise.
6373 (masked_increment): Likewise.
6374 (simplify_bit_ops_using_ranges): Likewise.
6376 2017-10-09 Martin Jambor <mjambor@suse.cz>
6379 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6380 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6381 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6383 (hsa_fixup_mov_insn_type): New function.
6384 (hsa_op_with_type::get_in_type): Use it.
6385 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6386 immediates in an assert.
6387 (hsa_op_with_type::extend_int_to_32bit): New method.
6388 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6389 types. Convert to dest type if necessary.
6390 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6391 (reg_for_gimple_ssa): Pass false as min32int to
6392 hsa_type_for_scalar_tree_type.
6393 (gen_hsa_addr): Fixup type when creating addresable temporary.
6394 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6395 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6396 necessary. Call hsa_fixup_mov_insn_type.
6397 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6398 extend operands and convert to dest type if necessary.
6399 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6400 to dest type if necessary.
6401 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6402 if conversion nt necessary and size matches.
6403 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6404 to dest type if necessary.
6405 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6406 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6408 (gen_hsa_clrsb): Likewise.
6409 (gen_hsa_ffs): Likewise.
6410 (gen_hsa_divmod): Extend operands and convert to dest type if
6412 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6414 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6416 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6417 Remove empty default arguments. Use a brace block as output
6419 (conditional return): Ditto.
6421 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6422 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6424 (group_ending_nop): Ditto.
6425 (doloop_end): Ditto.
6426 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6427 (splitters for those): Ditto.
6429 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6431 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6432 a conditional jump (and the compare for it) so that pc_rtx is the
6434 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6435 for the deleted and renamed ctr<mode>_internal[234] patterns.
6436 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6437 Delete second conditional return pattern.
6438 (ctr<mode>_internal2): Delete this second bdnz pattern.
6439 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6440 (ctr<mode>_internal4): Delete this second bdz pattern.
6442 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6444 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6445 (always_initialized_rtx_for_ssa_name_p): New predicate.
6446 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6447 (finish_out_of_ssa): Free new field of SA.
6448 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6449 * tree-ssa-coalesce.c: Include tree-ssa.h.
6450 (get_parm_default_def_partitions): Remove extern keyword.
6451 (get_undefined_value_partitions): New function.
6452 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6453 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6455 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6457 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6459 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6461 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6463 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6464 for targets that preffer 128bit.
6466 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6468 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6470 2017-10-08 Olivier Hainque <hainque@adacore.com>
6472 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6473 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6475 (thumb_set_return_address): Likewise.
6477 2017-10-08 Olivier Hainque <hainque@adacore.com>
6479 * common/config/arm/arm-common.c (arm_except_unwind_info):
6480 Handle DWARF2_UNWIND_INFO.
6482 2017-10-07 Michael Collison <michael.collison@arm.com>
6484 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6487 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6489 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6490 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6491 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6492 defined, force the creation of a new block for a dispatch label.
6494 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6496 * invoke.texi (Wsuggest-attribute=cold): Document.
6497 * common.opt (Wsuggest-attribute=cold): New
6498 * ipa-pure-const.c (warn_function_cold): New function.
6499 * predict.c (compute_function_frequency): Use it.
6500 * predict.h (warn_function_cold): Declare.
6502 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6504 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6507 2017-10-06 Martin Liska <mliska@suse.cz>
6509 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6510 keyword for member functions.
6511 (struct sanopt_tree_couple): New struct.
6512 (struct sanopt_tree_couple_hash): New function.
6513 (struct sanopt_ctx): Add new hash_map.
6514 (has_dominating_ubsan_ptr_check): New function.
6515 (record_ubsan_ptr_check_stmt): Likewise.
6516 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6517 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6518 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6520 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6523 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6524 aarch64_simd_valid_immediate on CONST_VECTORs.
6525 (aarch64_reg_or_bic_imm): Likewise.
6527 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6529 PR rtl-optimization/82396
6530 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6532 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6534 * graphite-dependences.c (scop_get_reads): Move code to...
6535 (scop_get_must_writes): Move code to...
6536 (scop_get_may_writes): Move code to...
6537 (scop_get_reads_and_writes): ... here.
6538 (scop_get_dependences): Call scop_get_reads_and_writes.
6540 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6542 PR tree-optimization/82434
6543 * fold-const.h (can_native_encode_type_p,
6544 can_native_encode_string_p): Remove.
6545 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6546 don't encode anything, just return what would be otherwise returned.
6547 (native_encode_fixed, native_encode_complex, native_encode_vector):
6549 (native_encode_string): Likewise. Inline by hand
6550 can_native_encode_string_p.
6551 (can_native_encode_type_p): Remove.
6552 (can_native_encode_string_p): Remove.
6553 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6554 STRING_CSTs using can_native_encode_string_p, test all
6555 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6556 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6557 argument from native_encode_expr.
6558 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6559 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6562 2017-10-06 Richard Biener <rguenther@suse.de>
6564 PR tree-optimization/82397
6565 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6566 operand_equal_p but rely on data_ref_compare_tree for detecting
6568 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6569 to match up with dr_group_sort_cmp.
6571 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6574 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6576 * config/s390/s390-builtin-types.def: Regenerate.
6578 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6581 * config/s390/s390-builtin-types.def: Regenerate.
6582 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6583 Change flag from B_VXE to B_VX.
6584 (s390_vec_min_dbl): Remove B_VXE flag.
6586 2017-10-06 Richard Biener <rguenther@suse.de>
6588 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6589 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6590 translate_isl_ast_to_gimple::is_valid_rename,
6591 translate_isl_ast_to_gimple::get_rename,
6592 translate_isl_ast_to_gimple::get_def_bb_for_const,
6593 translate_isl_ast_to_gimple::get_new_name,
6594 translate_isl_ast_to_gimple::collect_all_ssa_names,
6595 translate_isl_ast_to_gimple::copy_loop_phi_args,
6596 translate_isl_ast_to_gimple::collect_all_ssa_names,
6597 translate_isl_ast_to_gimple::copy_loop_phi_args,
6598 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6599 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6600 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6601 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6602 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6603 translate_isl_ast_to_gimple::copy_cond_phi_args,
6604 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6605 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6606 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6607 translate_isl_ast_to_gimple::rename_uses,
6608 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6609 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6610 (set_rename_for_each_def): Likewise.
6611 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6612 here. Handle rewriting SCEV analyzable uses here.
6613 (copy_bb_and_scalar_dependences): Generate code for PHI
6615 (graphite_regenerate_ast_isl): Adjust.
6616 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6617 (add_write, add_read): New functions.
6618 (build_cross_bb_scalars_def): Use it and simplify.
6619 (build_cross_bb_scalars_use): Likewise.
6620 (graphite_find_cross_bb_scalar_vars): Inline into...
6621 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6622 simulating out-of-SSA. Compute liveout and add dependencies.
6623 (build_scops): Force an empty entry block.
6624 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6626 (sese_build_liveouts): Declare.
6627 (sese_trivially_empty_bb_p): Likewise.
6628 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6629 compute liveout and debug_liveout.
6630 (sese_bad_liveouts_use): Remove.
6631 (sese_reset_debug_liveouts_bb): Likewise.
6632 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6633 (sese_build_liveouts): Build liveout and debug_liveout and store
6635 (new_sese_info): Adjust.
6636 (free_sese_info): Likewise.
6637 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6638 do not build liveout here.
6639 (move_sese_in_condition): Adjust region entry.
6640 (scev_analyzable_p): Match up with chrec_apply requirements.
6641 (sese_trivially_empty_bb_p): New.
6642 * tree-into-ssa.c (get_reaching_def): Properly support generating
6643 default-defs for incremental rewrite of anonymous names.
6645 2017-10-06 Richard Biener <rguenther@suse.de>
6647 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6648 precision do not perform modulo reduction.
6650 2017-10-06 Richard Biener <rguenther@suse.de>
6652 PR tree-optimization/82436
6653 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6654 conservatively choose the vectorization factor when checking
6655 whether we can perform the required load permutation.
6656 (vect_transform_slp_perm_load): Assert when we may not fail.
6658 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6660 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6661 message for incompatible -msdata=* and -mcall-* options.
6663 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6665 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6666 rate for post-reload scheduling.
6668 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6670 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6672 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6674 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6675 to improve monte carlo in scimark.
6677 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6679 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6680 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6681 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6682 pentium4_cost, nocona_cost): Set reassociation width to 1.
6683 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6684 width to 2 for fp operations and 1 otherwise.
6685 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6687 (atom_cost): Set reassociation width to 2.
6688 (slm_cost, generic_cost): Set fp reassociation width
6689 to 2 and 1 otherwise.
6690 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6691 (core_cost): Set fp reassociation width to 4 and vector to 2.
6692 (ix86_reassociation_width): Rewrite using cost table; special case
6693 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6694 and TARGET_AVX128_OPTIMAL.
6695 * config/i386/i386.h (processor_costs): Add
6696 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6697 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6698 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6699 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6700 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6701 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6703 2017-10-05 Nathan Sidwell <nathan@acm.org>
6705 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6707 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6709 * config/arm/arm.c (arm_test_fpu_data): New.
6710 (arm_run_selftests): Call arm_test_fpu_data.
6712 2017-10-04 Nathan Sidwell <nathan@acm.org>
6714 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6716 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6718 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6720 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6721 check type for aarch64_simd_valid_immediate.
6722 (aarch64_output_simd_mov_immediate): Update prototype.
6723 (aarch64_simd_valid_immediate): Update prototype.
6724 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6725 support for ORR-immediate.
6726 (and<mode>3): modified pattern to add support for BIC-immediate.
6727 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6728 now checks for valid immediate for BIC and ORR based on new enum
6730 (aarch64_output_simd_mov_immediate): Function now used to output
6731 BIC/ORR imm as well based on new enum argument.
6732 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6734 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6735 (aarch64_reg_or_bic_imm): Likewise.
6737 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6739 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6740 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6742 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6746 PR rtl-optimization/82396
6747 * haifa-sched.c (autopref_multipass_init): Simplify
6749 (autopref_rank_data): Simplify sort order.
6750 * sched-int.h (autopref_multipass_data_): Remove
6751 multi_mem_insn_p, min_offset and max_offset.
6753 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6755 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6757 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6759 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6762 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6764 * doc/sourcebuild.texi: Document vect_long_mult.
6766 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6768 PR tree-optimization/82413
6769 * fold-const.c (build_range_check): Use widest_int when comparing
6770 the maximum ETYPE value with HIGH.
6772 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6774 PR rtl-optimization/82396
6775 * haifa-sched.c (autopref_multipass_init): Simplify
6777 (autopref_rank_data): Simplify sort order.
6778 * sched-int.h (autopref_multipass_data_): Remove
6779 multi_mem_insn_p, min_offset and max_offset.
6781 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6783 PR tree-optimization/82381
6784 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6785 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6788 PR tree-optimization/82374
6789 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6790 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6791 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6792 current_function_decl to the new decl.
6794 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6796 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6797 helper macro for IEEE float128 hardware built-in functions.
6798 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6800 (TRUNCF128_ODD): Likewise.
6801 (ADDF128_ODD): Likewise.
6802 (SUBF128_ODD): Likewise.
6803 (MULF128_ODD): Likewise.
6804 (DIVF128_ODD): Likewise.
6805 (FMAF128_ODD): Likewise.
6806 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6807 UNSPEC_TRUNC_ROUND_TO_ODD.
6808 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6809 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6810 floating point round to odd instructions.
6811 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6812 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6813 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6814 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6815 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6816 (trunc<mode>sf2_hw): Change the truncate with round to odd
6817 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6818 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6819 to odd hardware instructions.
6820 (sub<mode>3_odd): Likewise.
6821 (mul<mode>3_odd): Likewise.
6822 (div<mode>3_odd): Likewise.
6823 (sqrt<mode>2_odd): Likewise.
6824 (fma<mode>4_odd): Likewise.
6825 (fms<mode>4_odd): Likewise.
6826 (nfma<mode>4_odd): Likewise.
6827 (nfms<mode>4_odd): Likewise.
6828 (trunc<mode>df2_odd): Change the truncate with round to odd
6829 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6831 * doc/extend.texi (PowerPC built-in functions): Update documentation
6832 for existing IEEE float128-bit built-in functions. Add built-in
6833 functions that generate the IEEE 128-bit floating point round to
6836 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6838 PR rtl-optimization/77729
6839 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6840 to (X&(C1&~C2))|C2 transformations.
6842 2017-10-03 Martin Jambor <mjambor@suse.cz>
6844 PR tree-optimization/82363
6845 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6846 mismatch, mark lacc written regardless of racc.
6848 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6850 PR tree-optimization/82381
6851 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6852 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6853 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6854 fallthrough into reassoc_stmt_dominates_stmt_p.
6857 * combine.c (combine_instructions): Don't combine in unreachable
6860 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6864 function to not use the have_cpu variable. Do not set cpu_index,
6865 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6867 (rs6000_valid_attribute_p): Remove duplicate initializations of
6868 old_optimize and func_optimize.
6869 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6870 (rs6000_activate_target_options): Make global.
6871 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6874 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6876 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6877 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6878 Set *poffset to 0 and *psize and *pmax_size to -1 if
6879 *poffset + *psize overflows in HOST_WIDE_INT.
6881 PR tree-optimization/82387
6882 PR tree-optimization/82388
6883 PR tree-optimization/82389
6884 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6885 instead of live_bytes non-NULL.
6887 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6890 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6893 2017-10-02 Richard Biener <rguenther@suse.de>
6895 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6896 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6897 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6899 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6901 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6902 requested precision matches the type's.
6903 * calls.c (alloc_max_size): Calculate the new candidate size as
6904 a widest_int and use wi::to_widest when comparing it with the
6905 current candidate size.
6906 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6907 zero rather than integer_zero_node.
6908 * match.pd: Check for a no-op conversion before using wi::add
6909 rather than after. Use tree_to_uhwi when summing small shift
6910 counts into an unsigned int.
6912 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6913 Alan Hayward <alan.hayward@arm.com>
6914 David Sherwood <david.sherwood@arm.com>
6917 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6918 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6919 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6920 POINTER_AND_FP_REGS.
6922 2017-10-02 Richard Biener <rguenther@suse.de>
6924 PR tree-optimization/82355
6925 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6926 a mapping for the enclosing loop but avoid generating one for
6928 (copy_bb_and_scalar_dependences): Remove premature codegen
6929 error on PHIs in blocks duplicated into multiple places.
6930 * graphite-scop-detection.c
6931 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6932 in the region use it as loop and nest to analyze the DR in.
6933 (try_generate_gimple_bb): Likewise.
6934 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6935 (add_loop_constraints): For blocks in a loop not in the region
6936 create a dimension with a single iteration.
6937 * sese.h (gbb_loop_at_index): Remove assert.
6939 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6941 * omp-expand.c (adjust_context_scope): New function.
6942 (expand_parallel_call): Call adjust_context_scope.
6944 2017-10-01 Jeff Law <law@redhat.com>
6946 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6947 dom_opt_dom_walker class with direct access to private members.
6948 Add comments. Call test_for_singularity.
6949 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6950 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6951 m_dummy_cond anymore.
6952 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6954 (pass_dominator:execute): Build the dummy_cond here and pass it
6955 to the dom_opt_dom_walker ctor.
6956 (test_for_singularity): New function.
6958 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6959 Maya Rashish <coypu@sdf.org>
6961 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6962 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6963 (alpha*-*-netbsd*) Use nbsd_tm_file.
6964 (arm*-*-netbsdelf*) Likewise.
6965 (i[34567]86-*-netbsdelf*) Likewise.
6966 (x86_64-*-netbsd*) Likewise.
6967 (mips*-*-netbsd*) Likewise.
6968 (powerpc-*-netbsd*) Likewise.
6969 (sh*-*-netbsd*) Likewise.
6970 (sparc-*-netbsdelf*) Likewise.
6971 (sparc64-*-netbsd*) Likewise.
6972 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6974 (vax-*-netbsdelf*) Likewise.
6975 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6976 (UINT_FAST8_TYPE) Likewise.
6977 (INT_FAST16_TYPE) Check CHAR_FAST16.
6978 (UINT_FAST16_TYPE) Likewise.
6980 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6983 * config/i386/i386.md
6984 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6985 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6986 *divmodsi4_zext_2): New define_insn_and_split.
6987 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6988 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6989 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6990 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6991 New define_insn_and_split.
6992 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6993 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6994 operands[1] having DImode when mode is SImode.
6996 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6997 always SImode for DIV and MOD in REG_EQUAL notes.
6999 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7002 * match.pd: Fix handling of NaNs in pattern.
7004 2017-09-29 Jeff Law <law@redhat.com>
7006 * sbitmap.c (bitmap_bit_in_range_p): New function.
7007 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7008 * tree-ssa-dse.c (live_bytes_read): New function.
7009 (dse_classify_store): Ignore reads of dead bytes.
7011 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7012 typos and whitespace errors.
7013 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7014 * config/s390/s390.c (s390_emit_prologue): Likewise.
7016 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7019 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7020 with a symbol for LRA.
7022 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7024 PR rtl-optimization/82338
7025 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7027 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7029 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7030 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7031 (qsort_chk): Declare.
7032 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7033 (qsort_chk): New function.
7035 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7037 PR tree-optimization/82337
7038 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7039 phi definition if the PHI result appears in an abnormal PHI.
7040 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7041 basis appears in an abnormal PHI.
7043 2017-09-29 Richard Biener <rguenther@suse.de>
7045 * graphite-isl-ast-to-gimple.c
7046 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7047 (binary_op_to_tree): Use it.
7048 (get_rename_from_scev): Likewise.
7049 (copy_loop_phi_nodes): Likewise.
7050 (copy_bb_and_scalar_dependences): Likewise.
7051 (translate_pending_phi_nodes): Likewise.
7053 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7056 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7057 for movabsq $(i32 << shift), r64.
7059 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7061 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7062 index when encoding %esp as %rsp to avoid 0x67 prefix.
7064 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7066 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7067 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7069 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7071 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7072 Extensions with more than 16 double VFP registers.
7073 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7074 to_clear_mask and all code related to it. Replace the remaining
7075 entry by a sbitmap and adapt code accordingly.
7077 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7079 * brig-builtins.def: Change pure attributes to const.
7081 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7083 * config.gcc (default_gnu_indirect_function): Default to yes for
7084 sparc*-*-linux* with glibc.
7086 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7088 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7089 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7090 when creating .init_array and .fini_array sections with priority
7093 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7096 * config/aarch64/aarch64.c
7097 (aarch64_builtin_support_vector_misalignment): Always return false
7098 when misalignment is unknown.
7100 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7102 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7103 this function to return false if the definition used by the swap
7104 instruction is artificial, or if the memory address from which the
7105 constant value is loaded is not represented by a base address held
7106 in a register or if the base address register is a frame or stack
7107 pointer. Additionally, return false if the base address of the
7108 loaded constant is a SYMBOL_REF but is not considered to be a
7110 (replace_swapped_load_constant): New function.
7111 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7112 loaded constant vector with a load of a swapped constant vector.
7114 2017-09-27 Carl Love <cel@us.ibm.com>
7116 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7117 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7118 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7121 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7123 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7124 first, always call autopref_rank_data otherwise.
7126 2017-09-27 Richard Biener <rguenther@suse.de>
7128 * graphite-scop-detection.c (find_scop_parameters): Move
7129 loop bound handling ...
7130 (gather_bbs::before_dom_children): ... here, avoiding the need
7131 to build scop_info->loop_nest.
7132 (record_loop_in_sese): Remove.
7133 * sese.h (sese_info_t::loop_nest): Remove.
7134 * sese.c (new_sese_info): Do not allocate loop_nest.
7135 (free_sese_info): Do not free loop_nest.
7137 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7140 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7141 lhs from calls if the lhs has addressable type.
7143 2017-09-27 Richard Biener <rguenther@suse.de>
7145 * graphite.h (scop::max_alias_set): New member.
7146 * graphite-scop-detection.c: Remove references to non-existing
7147 --param in comments.
7148 (build_alias_sets): Record the maximum alias set used for drs.
7149 (build_scops): Support zero as unlimited for
7150 --param graphite-max-arrays-per-scop.
7151 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7153 (build_poly_sr_1): ... here. Compute alias set based on the
7154 maximum alias set used for drs rather than
7155 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7157 2017-09-27 Richard Biener <rguenther@suse.de>
7159 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7160 --param loop-block-tile-size=0 to disable tiling.
7162 2017-09-27 Richard Biener <rguenther@suse.de>
7164 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7165 (graphite-max-nb-scop-params): Document special value zero.
7166 * domwalk.h (dom_walker::STOP): New symbolical constant.
7167 (dom_walker::dom_walker): Add optional parameter for bb to
7169 (dom_walker::~dom_walker): Declare.
7170 (dom_walker::before_dom_children): Document STOP return value.
7171 (dom_walker::m_user_bb_to_rpo): New member.
7172 (dom_walker::m_bb_to_rpo): Likewise.
7173 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7174 mapping here if not provided by the user.
7175 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7176 provided by the user.
7177 (dom_walker::STOP): Define.
7178 (dom_walker::walk): Do not compute bb to RPO mapping here.
7179 Support STOP return value from before_dom_children to stop
7181 * graphite-optimize-isl.c (optimize_isl): If the schedule
7182 is the same still generate code if -fgraphite-identity
7183 or -floop-parallelize-all are given.
7184 * graphite-scop-detection.c: Include cfganal.h.
7185 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7187 (gather_bbs::before_dom_children): Return STOP for BBs
7189 (build_scops): Compute bb to RPO mapping and pass it to
7190 the domwalk. Treat --param graphite-max-nb-scop-params=0
7191 as not limiting the number of params.
7192 * graphite.c (graphite_initialize): Remove limit on the
7193 number of basic-blocks in a function.
7194 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7195 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7196 default value of 10.
7198 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7200 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7201 Adjust code to eliminate needing to do the shift right 32-bits
7202 operation after XSCVDPSPN.
7204 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7206 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7207 ((X / Y) != 0 -> X >= Y): Likewise.
7209 2017-09-26 Carl Love <cel@us.ibm.com>
7211 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7212 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7213 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7214 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7215 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7216 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7217 definitions and overloading.
7218 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7219 statement for P9V_BUILTIN_XST_LEN_R.
7220 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7221 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7222 define_expand and define_insn for the instructions and builtins.
7223 * doc/extend.texi: Update the built-in documentation file for the new
7225 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7226 define_insn for the instructions
7228 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7231 * gcc/config/netbsd-protos.h: New file.
7232 * gcc/config/netbsd.c: New file.
7233 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7234 * gcc/config/t-netbsd: New file.
7235 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7236 (tmake_file) Add t-netbsd.
7237 (extra_objs) Add netbsd.o.
7239 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7243 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7245 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7247 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7248 sign extension from a vector register to a GPR by doing a 32-bit
7249 direct move and then an EXTSW.
7250 (extendsi<mode>2 splitter): Likewise.
7251 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7252 right or vector extract after doing XSCVDPSPN. Use
7253 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7255 (movdi_from_sf_zero_ext): Likewise.
7256 (reload_gpr_from_vsxsf): Likewise.
7257 (p8_mfvsrd_4_disf): Delete, no longer used.
7258 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7259 then needing to move the SFmode to a GPR to use the XSCVDPSP
7260 instruction instead of FRSP and XSCVDPSPN.
7261 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7262 it is adjacent to the other XSCVSPDP insns.
7263 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7264 SFmode to be in traditional Altivec registers.
7265 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7266 (vsx_xscvspdpn): Likewise.
7267 (vsx_xscvspdpn_scalar): Likewise.
7269 2017-09-26 Martin Jambor <mjambor@suse.cz>
7271 * tree-sra.c (compare_access_positions): Put integral types first,
7272 stabilize sorting of integral types, remove conditions putting
7273 non-full-precision integers last.
7274 (sort_and_splice_var_accesses): Disable scalarization if a
7275 non-integert would be represented by a non-full-precision integer.
7277 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7279 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7280 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7281 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7282 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7283 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7284 conditionals inside the function instead of around it. Call
7285 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7286 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7288 2017-09-26 Richard Biener <rguenther@suse.de>
7290 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7292 (scop_detection::build_scop_breadth): ... this. Removed.
7293 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7294 (scop_detection::harmful_stmt_in_bb): Likewise.
7295 (scop_detection::graphite_can_represent_stmt): Likewise.
7296 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7297 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7298 (scop_detection::can_represent_loop_1): ... this. Removed.
7299 (scop_detection::harmful_loop_in_region): Simplify after inlining
7300 the above and remove more quadraticness.
7301 (build_scops): Adjust.
7302 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7305 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7308 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7309 REGNO (base) == SP_REG if base is a REG.
7312 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7313 if it is different SSA_NAME.
7314 (optimize_range_tests_cmp_bitwise): New function.
7315 (optimize_range_tests): Call it.
7317 2017-09-26 Richard Biener <rguenther@suse.de>
7319 PR tree-optimization/82321
7320 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7321 for the def being inside the loop.
7323 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7325 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7327 * config/s390/s390-builtins.def: Fix constraint on op4.
7329 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7331 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7332 independent expanders.
7333 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7334 ("vec_ordered", "vec_unordered"): New expanders.
7336 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7338 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7341 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7343 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7344 vec_unpacks_lo_v16qi.
7345 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7347 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7349 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7350 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7351 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7353 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7355 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7357 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7359 ("*vec_slb<mode>"): New insn pattern.
7360 ("vec_shr_<mode>"): New expander.
7361 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7362 and force the shift count operand to V16QImode.
7363 ("vec_srb<mode>"): Set shift count mode to V16QI.
7365 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7367 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7368 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7369 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7371 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7374 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7376 2017-09-26 Richard Biener <rguenther@suse.de>
7378 PR tree-optimization/82320
7379 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7382 2017-09-25 Jeff Law <law@redhat.com>
7384 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7385 prototype for new argument.
7386 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7387 mostly extracted from rs6000_emit_allocate_stack.
7388 (rs6000_emit_probe_stack_range_stack_clash): New function.
7389 (rs6000_emit_allocate_stack): Call
7390 rs6000_emit_probe_stack_range_stack_clash as needed.
7391 (rs6000_emit_probe_stack_range): Add additional argument
7392 to call to gen_probe_stack_range{si,di}.
7393 (output_probe_stack_range): New.
7394 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7395 (output_probe_stack_range_stack_clash): New.
7396 (rs6000_emit_prologue): Emit notes into dump file as requested.
7397 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7398 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7399 Add additional operand and pass it to output_probe_stack_range.
7401 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7403 PR tree-optimization/82163
7404 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7405 (checking_verify_loop_closed_ssa): New parameter.
7406 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7407 (check_loop_closed_ssa_stmt): Delete.
7408 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7409 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7410 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7413 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7415 * brig-builtins.def: Treat HSAIL barrier builtins as
7416 setjmp/longjump style functions.
7418 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7420 * target.def (constant_alignment): New hook.
7421 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7422 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7423 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7424 * doc/tm.texi: Regenerate.
7425 * targhooks.h (default_constant_alignment): Declare.
7426 (constant_alignment_word_strings): Likewise.
7427 * targhooks.c (default_constant_alignment): New function.
7428 (constant_alignment_word_strings): Likewise.
7429 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7430 instead of CONSTANT_ALIGNMENT.
7431 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7432 (force_const_mem): Likewise.
7433 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7434 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7435 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7436 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7437 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7439 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7440 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7441 constant_alignment_word_strings.
7442 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7443 (CONSTANT_ALIGNMENT): Likewise.
7444 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7445 (arm_constant_alignment): New function.
7446 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7447 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7448 constant_alignment_word_strings.
7449 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7450 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7451 constant_alignment_word_strings.
7452 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7453 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7454 (cris_constant_alignment): New function.
7455 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7456 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7457 (epiphany_constant_alignment): New function.
7458 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7459 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7460 constant_alignment_word_strings.
7461 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7462 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7463 constant_alignment_word_strings.
7464 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7465 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7466 constant_alignment_word_strings.
7467 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7468 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7469 * config/i386/i386.c (ix86_constant_alignment): Make static.
7470 Use the same interface as the target hook.
7471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7472 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7473 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7474 constant_alignment_word_strings.
7475 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7476 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7477 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7478 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7479 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7480 constant_alignment_word_strings.
7481 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7482 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7483 constant_alignment_word_strings.
7484 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7485 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7486 constant_alignment_word_strings.
7487 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7488 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7490 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7491 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7492 * config/mips/mips.c (mips_constant_alignment): New function.
7493 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7494 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7495 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7496 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7497 (mmix_constant_alignment): Make static. Use the same interface
7499 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7500 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7501 constant_alignment_word_strings.
7502 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7503 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7504 constant_alignment_word_strings.
7505 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7506 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7507 constant_alignment_word_strings.
7508 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7509 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7510 (rs6000_constant_alignment): New function.
7511 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7512 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7513 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7514 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7515 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7516 (rs6000_constant_alignment): New function.
7517 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7518 * config/s390/s390.c (s390_constant_alignment): New function.
7519 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7520 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7521 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7522 constant_alignment_word_strings.
7523 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7524 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7525 (sparc_constant_alignment): New function.
7526 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7527 * config/spu/spu.c (spu_constant_alignment): New function.
7528 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7529 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7530 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7531 constant_alignment_word_strings.
7532 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7533 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7534 constant_alignment_word_strings.
7535 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7536 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7537 constant_alignment_word_strings.
7538 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7539 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7540 (visium_constant_alignment): New function.
7541 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7542 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7543 (xtensa_constant_alignment): New function.
7544 * system.h (CONSTANT_ALIGNMENT): Poison.
7546 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7549 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7550 (rs6000_builtin_valid_without_lhs): New helper function.
7551 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7552 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7554 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7556 * target.h (vec_perm_indices): Use unsigned short rather than
7558 (auto_vec_perm_indices): Likewise.
7559 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7560 Use unsigned int rather than unsigned char.
7561 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7563 2017-09-25 Richard Biener <rguenther@suse.de>
7565 * cfgloop.h (sort_sibling_loops): Declare.
7566 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7567 (sort_sibling_loops): New function sorting the sibling loop list
7569 * graphite.c (graphite_transform_loops): Sort sibling loops.
7571 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7573 * target.def (vec_perm_const_ok): Change sel parameter to
7575 * optabs-query.c (can_vec_perm_p): Update accordingly.
7576 * doc/tm.texi: Regenerate.
7577 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7578 auto_vec_perm_indices and remove separate nelt field.
7579 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7580 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7581 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7582 (aarch64_expand_vec_perm_const): Update accordingly.
7583 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7584 to vec_perm_indices.
7585 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7586 auto_vec_perm_indices and remove separate nelt field.
7587 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7588 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7589 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7591 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7592 to vec_perm_indices.
7593 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7594 sel to vec_perm_indices.
7595 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7596 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7597 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7599 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7602 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7605 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7606 on the FUNCTION_DECL function context if it has a DIE that is a
7609 2017-09-25 Richard Biener <rguenther@suse.de>
7611 PR tree-optimization/82285
7612 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7615 2017-09-25 Tom de Vries <tom@codesourcery.com>
7619 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7622 2017-09-25 Richard Biener <rguenther@suse.de>
7624 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7625 ISL errors other than isl_error_quota happen. Dump if the
7626 schedule is the same.
7627 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7628 errors instead of aborting inside ISL.
7630 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7633 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7634 of libgcc_eh for m64.
7635 * config/i386/darwin64.h: Likewise.
7637 2017-09-25 Richard Biener <rguenther@suse.de>
7640 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7641 attribute for incomplete types nor twice for complete ones.
7643 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7646 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7647 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7650 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7653 * config/i386/i386.opt (mprefer-avx256): Use
7654 ix86_target_flags variable.
7655 * config/i386/i386.c (ix86_target_string): Move
7656 -mprefer-avx256 to flag2_opts.
7658 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7661 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7662 and x != -1 | y != -1 into (x & y) != -1.
7664 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7666 * config.gcc: Add new case statement to set
7667 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7668 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7669 s390x-*-linux* case statements. Added aarch64 to the list of
7670 supported architectures.
7672 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7674 PR tree-optimization/82289
7675 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7676 STMT_VINFO_RELEVANT_P.
7678 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7679 Alan Hayward <alan.hayward@arm.com>
7680 David Sherwood <david.sherwood@arm.com>
7682 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7683 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7684 (extract_range_from_binary_expr_1): Don't call
7685 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7687 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7688 Alan Hayward <alan.hayward@arm.com>
7689 David Sherwood <david.sherwood@arm.com>
7691 * target.def (preferred_vector_alignment): New hook.
7692 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7694 * doc/tm.texi: Regenerate.
7695 * targhooks.h (default_preferred_vector_alignment): Declare.
7696 * targhooks.c (default_preferred_vector_alignment): New function.
7697 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7699 (DR_TARGET_ALIGNMENT): New macro.
7700 (aligned_access_p): Update commentary.
7701 (vect_known_alignment_in_bytes): New function.
7702 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7704 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7705 Calculate the misalignment based on the target alignment rather than
7707 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7708 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7709 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7710 the target alignment, rather than masking the element misalignment
7711 with the number of elements in a vector. Also use the target
7712 alignment when calculating the maximum number of peels.
7713 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7714 instead of TYPE_ALIGN_UNIT.
7715 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7716 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7717 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7718 (vect_create_data_ref_ptr): Likewise.
7719 (vect_setup_realignment): Realign by ANDing with
7720 -DR_TARGET_MISALIGNMENT.
7721 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7722 the number of peels based on DR_TARGET_ALIGNMENT.
7723 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7724 with the guaranteed alignment boundary when deciding whether
7726 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7727 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7728 (ensure_base_align): Remove stmt_info parameter. Get the
7729 target base alignment from DR_TARGET_ALIGNMENT.
7730 (vectorizable_store): Update call accordingly. Interpret
7731 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7733 (vectorizable_load): Likewise.
7735 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7736 Alan Hayward <alan.hayward@arm.com>
7737 David Sherwood <david.sherwood@arm.com>
7739 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7740 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7741 (vect_enhance_data_refs_alignment): Likewise.
7743 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7745 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7746 error. Only quit immediately if parsing is complete.
7747 (BEGIN): Initialize fatal_err and parse_done.
7748 (begin fpu, end fpu): Check number of arguments.
7749 (begin arch, end arch): Likewise.
7750 (begin cpu, end cpu): Likewise.
7751 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7752 (optalias): Likewise.
7754 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7756 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7757 * config/arm/arm-isa.h: Delete. Move definitions to ...
7758 * arm-cpus.in: ... here. Use new feature and fgroup values.
7759 * config/arm/arm.c (arm_option_override): Use lower case for feature
7761 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7762 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7763 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7765 (print_isa_bits_for): New function.
7766 (gen_isa): New function.
7767 (gen_comm_data): Use print_isa_bits_for.
7768 (define feature): New keyword.
7769 (define fgroup): New keyword.
7770 * config/arm/t-arm (TM_H): Remove.
7771 (GTM_H): Add arm-isa.h.
7772 (arm-isa.h): Add rule to generate file.
7773 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7774 case for feature bit names.
7776 2017-09-22 Richard Biener <rguenther@suse.de>
7778 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7780 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7781 print of no dependency loops ...
7782 * graphite.c (graphite_transform_loops): ... here.
7783 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7785 (same_close_phi_node, remove_duplicate_close_phi,
7786 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7787 (canonicalize_loop_closed_ssa): ... here and simplify.
7788 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7789 (optimize_isl): Use dump_printf_loc to tell when we stopped
7790 optimizing because of an ISL timeout.
7792 2017-09-22 Richard Biener <rguenther@suse.de>
7794 PR tree-optimization/82291
7795 * tree-if-conv.c (predicate_mem_writes): Make sure to
7796 remove writes in blocks predicated with false.
7798 2017-09-22 Richard Biener <rguenther@suse.de>
7800 * sese.c: Include cfganal.h.
7801 (if_region_set_false_region): Remove.
7802 (create_if_region_on_edge): Likewise.
7803 (move_sese_in_condition): Re-implement without destroying
7806 2017-09-22 Richard Biener <rguenther@suse.de>
7808 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7809 Verify both BBs contain loop PHI nodes before dispatching to
7811 (graphite_regenerate_ast_isl): Do not recompute dominators,
7812 do not verify three times. Restructure for clarity.
7813 * graphite-scop-detection.c (same_close_phi_node,
7814 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7815 defined_in_loop_p, canonicalize_loop_closed_ssa,
7816 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7817 checking and SSA rewrite, move to ...
7818 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7819 (graphite_initialize): Do not pass in ctx, do not reset the
7820 SCEV cache, compute only dominators.
7821 (graphite_transform_loops): Allocate ISL ctx after
7822 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7823 Maintain post-dominators only around build_scops.
7824 * sese.c (if_region_set_false_region): Make static. Free
7825 and recompute dominators.
7826 (move_sese_in_condition): Assert we don't get called with
7827 post-dominators computed.
7828 * sese.h (if_region_set_false_region): Remove.
7830 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7832 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7833 mode attribute for TARGET_AVX512VL.
7835 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7837 * config/i386/i386.opt (mprefer-avx256): New option.
7838 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7840 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7841 for TARGET_PREFER_AVX256.
7842 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7844 2017-09-21 Jeff Law <law@redhat.com>
7846 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7847 Fix dump output if the only stack space is for pushed registers.
7849 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7851 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7854 2017-09-21 Martin Sebor <msebor@redhat.com>
7857 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7858 code (in C++) or code that triggers warnings.
7860 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7862 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7864 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7866 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7867 * sched-rgn.c: Ditto.
7868 * sel-sched-ir.c: Ditto.
7870 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7872 * toplev.h (set_random_seed): Adjust return type.
7873 * toplev.c (init_local_tick): Move eager initialization of random_seed
7874 to get_random_seed. Adjust comment.
7875 (init_random_seed): Inline to get_random_seed, delete.
7876 (get_random_seed): Initialize random_seed lazily.
7877 (set_random_seed): Do not return previous value.
7878 (print_switch_value): Do not call get_random_seed.
7880 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7882 * cgraph.c (delete_function_version): New, broken out from...
7883 (cgraph_node::delete_function_version): ...here. Rename to
7884 cgraph_node::delete_function_version_by_decl. Update all uses.
7885 (cgraph_node::remove): Call delete_function_version.
7887 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7890 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7891 VAR_DECLs to which addressable non-volatile parameters are mapped
7892 and for id->retvar after the return value assignment. Clear
7893 id->retval and id->retbnd after inlining.
7895 2017-09-21 Richard Biener <rguenther@suse.de>
7897 PR tree-optimization/82276
7898 PR tree-optimization/82244
7899 * tree-vrp.c (build_assert_expr_for): Set
7900 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7902 (remove_range_assertions): Revert earlier change.
7904 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7907 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7909 2017-09-21 Richard Biener <rguenther@suse.de>
7911 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7912 Restore valid IL after code generation errors.
7913 * graphite.c (graphite_transform_loops): Diagnose code
7914 generation issues as MSG_MISSED_OPTIMIZATION and continue
7915 with processing SCOPs.
7917 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7918 Alan Hayward <alan.hayward@arm.com>
7919 David Sherwood <david.sherwood@arm.com>
7921 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7922 rather than choosing between plus_constant and gen_rtx_<CODE>.
7923 * expr.c (emit_push_insn): Likewise.
7924 (expand_expr_real_2): Likewise.
7926 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7927 Alan Hayward <alan.hayward@arm.com>
7928 David Sherwood <david.sherwood@arm.com>
7930 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7932 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7933 Alan Hayward <alan.hayward@arm.com>
7934 David Sherwood <david.sherwood@arm.com>
7936 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7937 calling tree_to_uhwi.
7939 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7940 Alan Hayward <alan.hayward@arm.com>
7941 David Sherwood <david.sherwood@arm.com>
7943 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7944 INTEGER_CST rather than a negative test for ADDR_EXPR.
7946 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7947 Alan Hayward <alan.hayward@arm.com>
7948 David Sherwood <david.sherwood@arm.com>
7950 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7951 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7953 2017-09-21 Richard Biener <rguenther@suse.de>
7955 PR tree-optimization/71351
7956 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7957 graphite_create_new_loop_guard): Remove, fold remaining parts
7959 (translate_isl_ast_node_for): ... here and simplify.
7961 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7964 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7965 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7966 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7967 alternative always use QI mode, for -Os imov (=R,R) alternative
7968 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7971 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7972 Jeff Law <law@redhat.com>
7974 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7975 (allocate_stack_space): New function, partially extracted from
7977 (s390_emit_prologue): Track offset to most recent stack probe.
7978 Code to allocate space moved into allocate_stack_space.
7979 Dump actions when no stack is allocated.
7980 (s390_prologue_plus_offset): New function.
7981 (s390_emit_stack_probe): Likewise.
7983 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7985 * common.opt (Wa, Wl, Wp, g, gz=): Add
7987 (gno-column-info): Remove.
7988 (gcolumn-info): Drop RejectNegative.
7990 (gno-record-gcc-switches): Remove.
7991 (grecord-gcc-switches): Drop RejectNegative.
7992 (gno-split-dwarf): Remove.
7993 (gsplit-dwarf): Drop RejectNegative.
7994 (gno-strict-dwarf): Remove.
7995 (gstrict-dwarf): Drop RejectNegative.
7996 * config/darwin.opt (gfull, gused): Add RejectNegative.
7997 * dwarf2out.c (gen_producer_string): Drop
7998 gno-record-gcc-switches handler.
7999 * optc-gen.awk: Add g to prefixes with negative forms.
8000 * opts-common.c (remapping_prefix_p): New.
8001 (find_opt): Check it.
8002 (generate_canonical_option): Test g prefix.
8003 (option_map): Add -gno- mapping.
8004 (add_misspelling_candidates): Check remapping_prefix_p.
8006 2017-09-20 Jeff Law <law@redhat.com>
8008 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8009 thinko in stack clash protection support.
8011 * explow.c (compute_stack_clash_protection_loop_data): Use
8012 CONST_INT_P instead of explicit test. Verify object is a
8013 CONST_INT_P before looking at INTVAL.
8014 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8015 instead of explicit test.
8017 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8020 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8021 address instead of to r1 and r11.
8023 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8025 * config.gcc: Support "knm".
8026 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8027 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8029 * config/i386/i386.c (m_KNM): Define.
8030 (processor_target_table): Add "knm".
8032 (ix86_option_override_internal): Add "knm".
8033 (ix86_issue_rate): Add PROCESSOR_KNM.
8034 (ix86_adjust_cost): Ditto.
8035 (ia32_multipass_dfa_lookahead): Ditto.
8036 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8037 (fold_builtin_cpu): Add M_INTEL_KNM.
8038 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8039 (processor_type): Add PROCESSOR_KNM.
8040 * config/i386/x86-tune.def: Add m_KNM.
8041 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8043 2017-09-20 Richard Biener <rguenther@suse.de>
8045 PR tree-optimization/80213
8046 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8047 are allowed in empty BBs as well.
8048 (canonicalize_loop_closed_ssa): Also look for other complex
8050 (scop_detection::get_sese): Include the loop-closed PHI block
8052 (scop_detection::merge_sese): Remove code adding extra blocks.
8053 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8054 (build_scops): Assert the final returned scop is invalid.
8056 2017-09-20 Richard Biener <rguenther@suse.de>
8058 PR tree-optimization/82264
8059 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8061 (vn_phi_lookup): Likewise.
8062 (vn_phi_insert): Likewise.
8064 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8066 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8067 that fits into uhwi or shwi, add DW_AT_const_value regardless
8068 of early_dwarf without going through RTL, using add_AT_unsigned
8071 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8072 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8073 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8074 (DEBUG_MACRO_SECTION): Likewise.
8075 (DEBUG_LTO_MACRO_SECTION): Likewise.
8076 (DEBUG_STR_DWO_SECTION): Likewise.
8077 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8078 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8079 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8080 (DEBUG_LTO_LINE_STR_SECTION): Define.
8081 (init_sections_and_labels): Initialize debug_line_str_section
8082 variable. Initialize debug_loc_section for -gdwarf-5 to
8083 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8085 2017-09-20 Richard Biener <rguenther@suse.de>
8087 * graphite-sese-to-poly.c (extract_affine): Properly handle
8088 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8090 2017-09-20 Richard Biener <rguenther@suse.de>
8092 PR tree-optimization/81373
8093 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8094 Force SESE live-out defs to be handled even if they are
8097 2017-09-19 Jeff Law <law@redhat.com>
8099 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8100 nothing for stack adjustments with REG_STACK_CHECK.
8101 * sched-deps.c (parse_add_or_inc): Reject insns with
8102 REG_STACK_CHECK from dependency breaking.
8103 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8104 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8105 * reg-notes.def (STACK_CHECK): New note.
8107 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8108 (ix86_expand_prologue): Dump stack clash info as needed.
8109 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8111 * function.c (dump_stack_clash_frame_info): New function.
8112 * function.h (dump_stack_clash_frame_info): Prototype.
8113 (enum stack_clash_probes): New enum.
8115 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8116 flag_stack_clash_protection.
8117 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8118 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8119 (arm_frame_pointer_required): Likewise.
8120 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8121 (ia64_expand_prologue): Likewise.
8122 * config/mips/mips.c (mips_expand_prologue): Likewise.
8123 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8124 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8125 (sparc_flat_expand_prologue): Likewise.
8126 * config/spu/spu.c (spu_expand_prologue): Likewise.
8128 * explow.c: Include "params.h".
8129 (anti_adjust_stack_and_probe_stack_clash): New function.
8130 (get_stack_check_protect): Likewise.
8131 (compute_stack_clash_protection_loop_data): Likewise.
8132 (emit_stack_clash_protection_loop_start): Likewise.
8133 (emit_stack_clash_protection_loop_end): Likewise.
8134 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8135 Use anti_adjust_stack_and_probe_stack_clash.
8136 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8137 (emit_stack_clash_protection_loop_start): Likewise.
8138 (emit_stack_clash_protection_loop_end): Likewise.
8139 * rtl.h (get_stack_check_protect): Prototype.
8140 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8141 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8142 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8144 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8146 * doc/tm.texi: Rebuilt.
8147 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8148 get_stack_check_protect.
8149 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8150 * config/arm/arm.c (arm_expand_prologue): Likewise.
8151 (arm_frame_pointer_required): Likewise.
8152 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8153 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8154 * config/mips/mips.c (mips_expand_prologue): Likewise.
8155 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8156 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8157 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8158 (sparc_flat_expand_prologue): Likewise.
8160 * common.opt (-fstack-clash-protection): New option.
8161 * flag-types.h (enum stack_check_type): Note difference between
8162 -fstack-check= and -fstack-clash-protection.
8163 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8164 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8165 * toplev.c (process_options): Issue warnings/errors for cases
8166 not handled with -fstack-clash-protection.
8167 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8168 (-fstack-check): Note additional problem with -fstack-check=generic.
8169 Note that -fstack-check is primarily for Ada and refer users
8170 to -fstack-clash-protection for stack-clash-protection.
8171 Document new params for stack clash protection.
8173 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8175 * config/i386/i386.c (ix86_split_long_move): Do not handle
8176 address used for LEA in a special way.
8178 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8180 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8182 2017-09-19 Martin Sebor <msebor@redhat.com>
8185 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8186 of incompatible types.
8188 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8190 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8191 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8192 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8193 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8195 2017-09-19 Richard Biener <rguenther@suse.de>
8197 PR tree-optimization/82244
8198 * tree-vrp.c (remove_range_assertions): Do not propagate
8199 a constant to abnormals but replace the assert with a copy.
8201 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8203 PR rtl-optimization/57878
8204 PR rtl-optimization/68988
8205 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8206 avoidance test involving non_reload_pseudos. Move frequency test
8207 below the general fragmentation avoidance test.
8209 2017-09-19 Richard Biener <rguenther@suse.de>
8211 PR tree-optimization/69728
8212 * graphite-sese-to-poly.c (schedule_error): New global.
8213 (add_loop_schedule): Handle empty domain by failing the
8215 (build_original_schedule): Handle schedule_error.
8217 2017-09-19 Richard Biener <rguenther@suse.de>
8219 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8220 Do not iterate to sibling loops but only to siblings of inner
8223 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8226 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8227 that the registers are different.
8229 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8231 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8232 to processor_model and "amdfam17h" to arch_names_table.
8233 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8235 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8238 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8240 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8241 Alan Hayward <alan.hayward@arm.com>
8242 David Sherwood <david.sherwood@arm.com>
8244 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8246 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8248 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8249 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8250 vect_schedule_slp_instance.
8251 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8252 Update call to vect_slp_analyze_node_operations. Simplify return
8254 (vect_slp_analyze_bb_1): Update call accordingly.
8255 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8256 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8257 (vect_schedule_slp): Update call accordingly.
8259 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8264 with types that aren't in fact scalar.
8266 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8268 * tree-vect-slp.c (vect_record_max_nunits): New function,
8270 (vect_build_slp_tree_1): ...here.
8271 (vect_build_slp_tree_2): Call it for phis too.
8273 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8275 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8276 to vect_get_vec_def_for_operand when getting the mask operand.
8278 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8279 Alan Hayward <alan.hayward@arm.com>
8280 David Sherwood <david.sherwood@arm.com>
8282 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8285 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8286 Alan Hayward <alan.hayward@arm.com>
8287 David Sherwood <david.sherwood@arm.com>
8289 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8290 calculation for vector booleans.
8292 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8293 Alan Hayward <alan.hayward@arm.com>
8294 David Sherwood <david.sherwood@arm.com>
8296 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8298 (vect_transform_stmt): ...here.
8299 (vect_analyze_stmt): Use it instead of calling
8300 vectorizable_live_operation directly.
8302 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8304 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8305 non-SIMT targets in acc vector loops.
8307 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8309 * configure.ac: Add arc and check if assembler supports gdwarf2.
8310 * configure: Regenerate.
8312 2017-09-18 Richard Biener <rguenther@suse.de>
8314 PR tree-optimization/82220
8315 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8316 epilogue niters from the min_profitable_iters compute.
8318 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8321 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8322 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8323 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8325 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8328 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8329 switching to a new text section.
8331 2017-09-18 Richard Biener <rguenther@suse.de>
8333 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8335 (build_alias_set): Reject aliases with no access function.
8337 2017-09-18 Richard Biener <rguenther@suse.de>
8339 PR tree-optimization/79622
8340 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8342 (build_cross_bb_scalars_use): Likewise.
8344 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8346 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8348 2017-09-18 Alan Modra <amodra@gmail.com>
8351 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8352 stack_pointer_rtx for count 0. Update comments. Break up
8353 large rtl expression.
8355 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8357 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8358 Increase to 20 bytes.
8359 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8360 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8361 or avx version of the stub.
8363 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8366 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8367 compute the minimum stack alignment. Also update preferred stack
8368 boundary for leaf functions.
8370 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8372 PR tree-optimization/82228
8373 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8376 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8378 * common/config/nds32/nds32-common.c
8379 (nds32_option_optimization_table): Refine formatting.
8380 (nds32_option_optimization_table): Use -fsched-pressure and
8381 -fomit-frame-pointer for specific optimization level.
8383 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8385 * config/nds32/nds32.c: Refine formatting and comments.
8386 * config/nds32/nds32.h: Likewise.
8387 * config/nds32/nds32.md: Likewise.
8388 * config/nds32/nds32-cost.c: Likewise.
8389 * config/nds32/nds32-isr.c: Likewise.
8390 * config/nds32/nds32-md-auxiliary.c: Likewise.
8391 * config/nds32/nds32-multiple.md: Likewise.
8392 * config/nds32/nds32-predicates.c: Likewise.
8394 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8395 Jakub Jelinek <jakub@redhat.com>
8397 Add support for -std=c++2a.
8398 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8400 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8402 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8405 * doc/extend.texi (Common Function Attributes): Add
8406 references to ARM, AArch64, and S/390 specific attributes.
8407 (Function Specific Option Pragmas): Add AArch64 and S/390
8408 to list of back ends that support the target pragma.
8410 2017-09-15 Nathan Sidwell <nathan@acm.org>
8412 * doc/standards.texi: Fix C++17 description. Update URLs for
8415 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8417 * common.opt (Wcast-align=strict): New warning option.
8418 * doc/invoke.texi: Document -Wcast-align=strict.
8420 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8422 * cgraph.h (cgraph_thunk_info): Add comments.
8423 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8424 assert for VIRTUAL_* arguments stricter.
8426 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8428 PR tree-optimization/71026
8429 * match.pd: Move RDIV patterns from fold-const.c
8430 * fold-const.c (distribute_real_division): Removed.
8431 (fold_binary_loc): Remove calls to distribute_real_divison.
8433 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8435 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8436 c++1z and gnu++1z as deprecated. Change other references to
8437 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8438 Change -Wc++1z-compat to -Wc++17-compat.
8439 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8440 * dwarf2out.c (highest_c_language): Handle C++17.
8441 (gen_compile_unit_die): Likewise.
8443 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8445 PR rtl-optimization/82192
8446 * combine.c (make_extraction): Don't look through non-paradoxical
8447 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8450 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8451 Alan Hayward <alan.hayward@arm.com>
8452 David Sherwood <david.sherwood@arm.com>
8454 * target.def (function_arg_offset): New hook.
8455 * targhooks.h (default_function_arg_offset): Declare.
8456 * targhooks.c (default_function_arg_offset): New function.
8457 * function.c (locate_and_pad_parm): Use
8458 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8459 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8460 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8461 * doc/tm.texi: Regenerate.
8462 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8463 * config/spu/spu.c (spu_function_arg_offset): New function.
8464 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8465 * system.h (FUNCTION_ARG_OFFSET): Poison.
8467 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8468 Alan Hayard <alan.hayward@arm.com>
8469 David Sherwood <david.sherwood@arm.com>
8471 * target.def (truly_noop_truncation): New hook.
8472 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8473 than TRULY_NOOP_TRUNCATION.
8474 * hooks.h (hook_bool_uint_uint_true): Declare.
8475 * hooks.c (hook_bool_uint_uint_true): New function.
8476 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8477 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8478 * doc/tm.texi: Regenerate.
8479 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8480 rather than TRULY_NOOP_TRUNCATION in comments.
8481 (simplify_comparison): Likewise.
8482 (record_truncated_value): Likewise.
8483 * expmed.c (extract_bit_field_1): Likewise.
8484 (extract_split_bit_field): Likewise.
8485 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8486 instead of TRULY_NOOP_TRUNCATION.
8487 * function.c (assign_parm_setup_block): Likewise.
8488 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8489 * rtlhooks.c: Include target.h.
8490 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8491 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8492 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8493 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8494 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8495 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8496 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8497 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8498 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8499 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8500 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8501 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8502 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8503 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8504 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8505 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8506 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8507 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8508 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8509 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8510 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8511 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8512 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8513 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8514 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8515 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8516 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8517 rather than TRULY_NOOP_TRUNCATION in comments.
8518 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8519 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8520 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8521 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8522 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8523 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8524 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8525 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8526 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8527 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8528 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8529 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8530 rather than TRULY_NOOP_TRUNCATION in comments.
8531 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8532 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8533 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8534 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8535 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8536 TRULY_NOOP_TRUNCATION condition.
8537 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8538 (TRULY_NOOP_TRUNCATION): Delete.
8539 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8540 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8541 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8542 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8543 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8544 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8545 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8546 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8547 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8548 rather than TRULY_NOOP_TRUNCATION in comments.
8549 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8550 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8551 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8552 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8553 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8554 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8556 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8559 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8560 (*cmp_ior): Likewise.
8561 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8562 (*ior_scc_scc_cmp): Likewise.
8563 (*and_scc_scc): Likewise.
8564 (*and_scc_scc_cmp): Likewise.
8566 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8567 Alan Hayard <alan.hayward@arm.com>
8568 David Sherwood <david.sherwood@arm.com>
8570 * target.def (can_change_mode_class): New hook.
8571 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8572 (hard_regno_nregs): Likewise.
8573 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8574 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8575 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8576 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8577 (LOAD_EXTEND_OP): Update accordingly.
8578 * doc/tm.texi: Regenerate.
8579 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8580 CANNOT_CHANGE_MODE_CLASS.
8581 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8582 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8583 * combine.c (simplify_set): Update accordingly.
8584 * emit-rtl.c (validate_subreg): Likewise.
8585 * recog.c (general_operand): Likewise.
8586 * regcprop.c (mode_change_ok): Likewise.
8587 * reload1.c (choose_reload_regs): Likewise.
8588 (inherit_piecemeal_p): Likewise.
8589 * rtlanal.c (simplify_subreg_regno): Likewise.
8590 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8591 instead of CANNOT_CHANGE_MODE_CLASS.
8592 (reload_cse_simplify_operands): Likewise.
8593 * reload.c (push_reload): Use targetm.can_change_mode_class
8594 instead of CANNOT_CHANGE_MODE_CLASS.
8595 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8596 REG_CANNOT_CHANGE_MODE_P.
8597 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8598 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8599 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8600 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8601 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8602 (arm_can_change_mode_class): New function.
8603 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8604 than CANNOT_CHANGE_MODE_CLASS in comments.
8605 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8606 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8607 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8608 (ix86_can_change_mode_class): ...this new function, inverting the
8609 sense of the return value.
8610 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8611 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8612 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8613 (ia64_can_change_mode_class): New function.
8614 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8615 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8616 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8617 (m32c_can_change_mode_class): ...this new function, inverting the
8618 sense of the return value.
8619 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8620 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8621 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8622 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8623 (mips_can_change_mode_class): ...this new function, inverting the
8624 sense of the return value.
8625 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8626 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8627 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8628 (msp430_can_change_mode_class): New function.
8629 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8630 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8631 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8632 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8633 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8634 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8635 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8636 (pa_cannot_change_mode_class): Replace with...
8637 (pa_can_change_mode_class): ...this new function, inverting the
8638 sense of the return value.
8639 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8640 than CANNOT_CHANGE_MODE_CLASS in comments.
8641 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8642 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8643 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8644 (pdp11_cannot_change_mode_class): Replace with...
8645 (pdp11_can_change_mode_class): ...this new function, inverting the
8646 sense of the return value.
8647 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8648 * config/powerpcspe/powerpcspe-protos.h
8649 (rs6000_cannot_change_mode_class_ptr): Delete.
8650 * config/powerpcspe/powerpcspe.c
8651 (rs6000_cannot_change_mode_class_ptr): Delete.
8652 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8653 (rs6000_option_override_internal): Assign to
8654 targetm.can_change_mode_class instead of
8655 rs6000_cannot_change_mode_class_ptr.
8656 (rs6000_cannot_change_mode_class): Replace with...
8657 (rs6000_can_change_mode_class): ...this new function, inverting the
8658 sense of the return value.
8659 (rs6000_debug_cannot_change_mode_class): Replace with...
8660 (rs6000_debug_can_change_mode_class): ...this new function.
8661 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8662 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8663 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8664 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8665 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8667 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8668 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8669 (rs6000_option_override_internal): Assign to
8670 targetm.can_change_mode_class instead of
8671 rs6000_cannot_change_mode_class_ptr.
8672 (rs6000_cannot_change_mode_class): Replace with...
8673 (rs6000_can_change_mode_class): ...this new function, inverting the
8674 sense of the return value.
8675 (rs6000_debug_cannot_change_mode_class): Replace with...
8676 (rs6000_debug_can_change_mode_class): ...this new function.
8677 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8678 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8679 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8680 (s390_can_change_mode_class): ...this new function, inverting the
8681 sense of the return value.
8682 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8683 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8684 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8685 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8686 (sh_cannot_change_mode_class): Replace with...
8687 (sh_can_change_mode_class): ...this new function, inverting the
8688 sense of the return value.
8689 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8690 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8691 (sparc_can_change_mode_class): New function.
8692 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8693 * config/spu/spu.c (spu_can_change_mode_class): New function.
8694 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8695 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8696 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8697 (visium_can_change_mode_class): New function.
8698 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8700 2017-09-15 Richard Biener <rguenther@suse.de>
8702 PR tree-optimization/82217
8703 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8704 but not undefined case.
8706 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8709 * postreload.c (reload_cse_simplify_operands): Skip
8710 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8712 2017-09-15 Richard Biener <rguenther@suse.de>
8714 PR tree-optimization/68823
8715 * graphite-scop-detection.c (build_alias_set): If we have a
8716 possible dependence check whether we can handle them by just
8717 looking at the DRs DR_ACCESS_FNs.
8718 (build_scops): If build_alias_set fails, fail the SCOP.
8720 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8722 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8723 to support float128 built-in functions that require the ISA 3.0
8725 (BU_FLOAT128_3_HW): Likewise.
8726 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8728 (FMAF128): Likewise.
8730 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8731 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8732 floating point instructions.
8733 (rs6000_invalid_builtin): Likewise.
8734 (rs6000_builtin_mask_names): Likewise.
8735 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8736 (RS6000_BTM_FLOAT128_HW): Likewise.
8737 (RS6000_BTM_COMMON): Likewise.
8738 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8740 * doc/extend.texi (RS/6000 built-in functions): Document the
8741 IEEE 128-bit floating point square root and fused multiply-add
8744 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8746 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8747 reg (r2) isn't in the set of registers defined in the prologue.
8749 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8750 Alan Hayward <alan.hayward@arm.com>
8751 David Sherwood <david.sherwood@arm.com>
8753 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8754 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8755 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8756 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8757 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8759 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8760 max_vectorization_factor.
8761 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8763 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8764 Alan Hayward <alan.hayward@arm.com>
8765 David Sherwood <david.sherwood@arm.com>
8767 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8768 (vect_worthwhile_without_simd_p): Declare.
8769 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8770 (vectorizable_reduction): Use it.
8771 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8772 (vectorizable_operation): Likewise.
8774 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8775 Alan Hayward <alan.hayward@arm.com>
8776 David Sherwood <david.sherwood@arm.com>
8778 * tree-vectorizer.h (vect_get_num_copies): New function.
8779 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8780 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8781 (vectorizable_induction): Likewise.
8782 (vectorizable_live_operation): Likewise.
8783 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8784 (vectorizable_bswap): Likewise.
8785 (vectorizable_call): Likewise.
8786 (vectorizable_conversion): Likewise.
8787 (vectorizable_assignment): Likewise.
8788 (vectorizable_shift): Likewise.
8789 (vectorizable_operation): Likewise.
8790 (vectorizable_store): Likewise.
8791 (vectorizable_load): Likewise.
8792 (vectorizable_condition): Likewise.
8793 (vectorizable_comparison): Likewise.
8794 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8796 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8797 Alan Hayward <alan.hayward@arm.com>
8798 David Sherwood <david.sherwood@arm.com>
8800 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8801 of vect_init_vector.
8803 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8804 Alan Hayward <alan.hayward@arm.com>
8805 David Sherwood <david.sherwood@arm.com>
8807 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8808 an inline wrapper that provides a location.
8809 (gimple_build_vector): Likewise.
8810 * gimple-fold.c (gimple_build_vector_from_val): New function.
8811 (gimple_build_vector): Likewise.
8812 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8813 functions to build the initial value. Always return a gimple value.
8814 (get_initial_defs_for_reduction): Likewise. Only compute
8816 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8817 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8818 (vectorizable_induction): Use gimple_build_vector rather than
8821 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8822 Alan Hayward <alan.hayward@arm.com>
8823 David Sherwood <david.sherwood@arm.com>
8825 * target.h (vec_perm_indices): New typedef.
8826 (auto_vec_perm_indices): Likewise.
8827 * optabs-query.h: Include target.h
8828 (can_vec_perm_p): Take a vec_perm_indices *.
8829 * optabs-query.c (can_vec_perm_p): Likewise.
8830 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8831 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8832 * tree-vect-generic.c (lower_vec_perm): Likewise.
8833 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8834 (vect_grouped_load_supported): Likewise.
8835 (vect_shift_permute_load_chain): Likewise.
8836 (vect_permute_store_chain): Use auto_vec_perm_indices.
8837 (vect_permute_load_chain): Likewise.
8838 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8839 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8840 Update uses of can_vec_perm_p.
8841 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8842 mode with a number of elements. Take a vec_perm_indices *.
8843 (vect_create_epilog_for_reduction): Update accordingly.
8844 Use auto_vec_perm_indices.
8845 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8846 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8847 (vect_transform_slp_perm_load): Likewise.
8848 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8849 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8850 (vect_gen_perm_mask_checked): Likewise.
8851 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8852 (vect_gen_perm_mask_checked): Likewise.
8853 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8854 (vectorizable_store): Likewise.
8855 (vectorizable_load): Likewise.
8856 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8857 (vectorizable_bswap): Likewise.
8859 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8860 Alan Hayward <alan.hayward@arm.com>
8861 David Sherwood <david.sherwood@arm.com>
8863 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8864 * tree.c (build_vector): Likewise.
8865 (build_vector_from_ctor): Update accordingly.
8866 (build_vector_from_val): Likewise.
8867 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8868 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8869 * tree-vect-generic.c (add_rshift): Likewise.
8870 (expand_vector_divmod): Likewise.
8871 (optimize_vector_constructor): Likewise.
8872 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8873 (vect_transform_slp_perm_load): Likewise.
8874 (vect_schedule_slp_instance): Likewise.
8875 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8876 (vectorizable_call): Likewise.
8877 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8878 * expmed.c (make_tree): Likewise.
8879 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8880 a vector passed to build_vector.
8881 (fold_convert_const): Likewise.
8882 (exact_inverse): Likewise.
8883 (fold_ternary_loc): Likewise.
8884 (fold_relational_const): Likewise.
8885 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8886 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8887 (const_unop): Likewise. Store the reduction accumulator in a
8888 variable rather than an array.
8889 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8890 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8891 the new vector, rather than constructing it after the input arrays.
8892 (native_interpret_vector): Use auto_vec<tree> when building
8893 a vector passed to build_vector. Add elements in order.
8894 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8895 auto_vec<tree> when building a vector passed to build_vector.
8896 (vect_create_epilog_for_reduction): Likewise.
8897 (vectorizable_induction): Likewise.
8898 (get_initial_def_for_reduction): Likewise. Fix indentation of
8900 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8902 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8903 passed to build_vector.
8905 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8906 Alan Hayward <alan.hayward@arm.com>
8907 David Sherwood <david.sherwood@arm.com>
8909 * tree-core.h (tree_base::u): Add an "nelts" field.
8910 (tree_vector): Use VECTOR_CST_NELTS as the length.
8911 * tree.c (tree_size): Likewise.
8912 (make_vector): Initialize VECTOR_CST_NELTS.
8913 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8914 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8915 TYPE_VECTOR_SUBPARTS.
8916 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8917 as the number of units, setting it from VECTOR_CST_NELTS.
8918 (const_vector_from_tree): Likewise.
8919 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8920 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8921 (fold_negate_expr_1): Likewise.
8922 (fold_convert_const): Likewise.
8923 (const_binop): Likewise. Differentiate the number of output and
8925 (const_unop): Likewise.
8926 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8927 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8928 in cases that did the opposite.
8930 2017-09-14 Richard Biener <rguenther@suse.de>
8932 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8935 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8937 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8939 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8942 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8943 if and where to split a bb, except for splitting before debug insn
8944 sequences followed by non-label real insn. Delete debug insns
8945 in between basic blocks.
8947 * combine.c (make_compound_operation_int): Formatting fixes.
8949 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8950 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8951 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8952 * config/sol2.h (LINK_EH_SPEC): Likewise.
8953 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8954 * config/s390/linux.h (LINK_SPEC): Likewise.
8955 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8956 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8957 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8958 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8959 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8960 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8961 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8962 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8964 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8966 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8968 (ENDFILE_LINUX_SPEC): Likewise.
8969 (LINK_EH_SPEC): Likewise.
8970 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8971 (LINK_OS_LINUX_SPEC32): Likewise.
8972 (LINK_OS_LINUX_SPEC64): Likewise.
8973 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8974 (LINK_OS_LINUX_SPEC): Likewise.
8976 2017-09-13 Martin Liska <mliska@suse.cz>
8979 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8980 CASE_HIGH is NULL_TREE.
8982 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8983 Alan Hayward <alan.hayward@arm.com>
8984 David Sherwood <david.sherwood@arm.com>
8986 * target.def (secondary_memory_needed): New hook.
8987 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8988 instead of SECONDARY_MEMORY_NEEDED.
8989 (secondary_memory_needed_mode): Likewise.
8990 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8991 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8992 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8993 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8994 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8995 * doc/tm.texi: Regenerate.
8996 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8997 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8998 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8999 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9000 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9001 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9002 mode argument first and change the reg_class arguments to reg_class_t.
9003 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9004 Make static. Update the call to inline_secondary_memory_needed.
9005 (ix86_register_move_cost): Update the call to
9006 inline_secondary_memory_needed.
9007 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9008 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9010 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9011 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9013 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9014 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9015 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9016 and match hook interface. Add comment from mips.h.
9017 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9018 * config/mmix/mmix.md (truncdfsf2): Refer to
9019 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9021 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9022 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9023 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9024 (pa_secondary_memory_needed): New function.
9025 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9026 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9027 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9028 (pdp11_secondary_memory_needed): Make static and match hook interface.
9029 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9030 * config/powerpcspe/powerpcspe-protos.h
9031 (rs6000_secondary_memory_needed_ptr): Delete.
9032 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9034 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9035 (rs6000_option_override_internal): Assign to
9036 targetm.secondary_memory_needed rather than
9037 rs6000_secondary_memory_needed_ptr.
9038 (rs6000_secondary_memory_needed): Match hook interface.
9039 (rs6000_debug_secondary_memory_needed): Likewise.
9040 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9041 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9042 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9043 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9044 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9045 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9047 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9048 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9049 (rs6000_option_override_internal): Assign to
9050 targetm.secondary_memory_needed rather than
9051 rs6000_secondary_memory_needed_ptr.
9052 (rs6000_secondary_memory_needed): Match hook interface.
9053 (rs6000_debug_secondary_memory_needed): Likewise.
9054 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9055 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9056 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9057 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9058 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9059 (sparc_secondary_memory_needed): New function.
9060 * lra-constraints.c (check_and_process_move): Refer to
9061 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9063 (curr_insn_transform): Likewise.
9064 (process_alt_operands): Use targetm.secondary_memory_needed
9065 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9066 (check_secondary_memory_needed_p): Likewise.
9067 (choose_split_class): Likewise.
9068 * reload.c: Unconditionally include code that was previously
9069 conditional on SECONDARY_MEMORY_NEEDED.
9070 (push_secondary_reload): Use targetm.secondary_memory_needed
9071 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9072 (push_reload): Likewise.
9073 * reload1.c: Unconditionally include code that was previously
9074 conditional on SECONDARY_MEMORY_NEEDED.
9075 (choose_reload_regs): Use targetm.secondary_memory_needed
9076 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9077 (gen_reload): Likewise.
9078 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9080 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9081 Alan Hayward <alan.hayward@arm.com>
9082 David Sherwood <david.sherwood@arm.com>
9084 * target.def (secondary_memory_needed_mode): New hook:
9085 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9086 * targhooks.h (default_secondary_memory_needed_mode): New function.
9087 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9088 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9089 * doc/tm.texi: Regenerate.
9090 * lra-constraints.c (check_and_process_move): Use
9091 targetm.secondary_memory_needed_mode instead of
9092 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9093 (curr_insn_transform): Likewise.
9094 * reload.c (get_secondary_mem): Likewise.
9095 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9096 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9098 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9099 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9100 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9101 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9102 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9104 * config/powerpcspe/powerpcspe-protos.h
9105 (rs6000_secondary_memory_needed_mode): Delete.
9106 * config/powerpcspe/powerpcspe.c
9107 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9108 (rs6000_secondary_memory_needed_mode): Make static.
9109 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9110 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9112 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9114 (rs6000_secondary_memory_needed_mode): Make static.
9115 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9116 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9117 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9118 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9119 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9121 (sparc_secondary_memory_needed_mode): New function.
9122 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9124 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9126 * config/aarch64/constraints.md (Umq): New constraint.
9127 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9129 (mov<mode>): Update condition.
9131 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9133 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9134 when bitposition is the same.
9136 2017-09-13 Richard Biener <rguenther@suse.de>
9138 * dwarf2out.c (output_die_symbol): Remove.
9139 (output_die): Do not output a DIEs symbol.
9141 2017-09-13 Richard Biener <rguenther@suse.de>
9144 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9145 default-def to avoid breaking iterator update with the weird
9146 interaction with cgraph_update_edges_for_call_stmt_node.
9148 2017-09-13 Richard Biener <rguenther@suse.de>
9150 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9151 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9152 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9153 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9155 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9157 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9158 Disable pc relative literal load irrespective of
9159 TARGET_FIX_ERR_A53_84341 for default.
9161 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9163 * config/sparc/sparc.c (output_return): Output the source location of
9164 the insn in the delay slot, if any.
9165 (output_sibcall): Likewise.
9167 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9170 * common.opt (-static-pie): New alias.
9171 (shared): Negate static-pie.
9172 (-no-pie): Update help text.
9174 (static-pie): New option.
9175 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9176 -static-pie support.
9177 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9178 (LINK_EH_SPEC): Likewise.
9179 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9180 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9181 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9182 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9183 (init_gcc_specs): Likewise.
9184 (init_spec): Likewise.
9185 (display_help): Update help message for -pie.
9186 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9189 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9191 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9192 (movdi_aarch64): Likewise.
9193 (movti_aarch64): Likewise.
9195 2017-09-12 Simon Wright <simon@pushface.org>
9198 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9199 calculation of the minor version, always output as 0.
9201 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9204 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9205 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9206 on it early, rather than manual conversion late. For
9207 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9208 instead of performing manual conversion.
9210 2017-09-12 Carl Love <cel@us.ibm.com>
9212 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9213 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9215 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9216 VMULOSW): Add definitions.
9217 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9218 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9219 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9220 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9221 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9223 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9225 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9227 (movti_aarch64): Likewise.
9228 (movdf_aarch64): Likewise.
9229 (movtf_aarch64): Likewise.
9230 (load_pairdi): Likewise.
9231 (store_pairdi): Likewise.
9232 (load_pairdf): Likewise.
9233 (store_pairdf): Likewise.
9234 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9235 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9236 (ldr_got_small_<mode>): Likewise.
9237 (ldr_got_small_28k_<mode>): Likewise.
9238 (ldr_got_tiny): Likewise.
9239 * config/aarch64/iterators.md (ldst_sz): New.
9240 (ldpstp_sz): Likewise.
9241 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9243 (thunderx_load): Split load_8 to load_16.
9244 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9246 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9247 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9248 (xgene1_store_pair): Split store_8 to store_16.
9249 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9250 (falkor_st_0_st_sd): Split store_8 to store_16.
9252 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9254 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9255 and store1/2/3/4 to store_4/8/12/16.
9256 * config/aarch64/aarch64.md: Update for rename.
9257 * config/arm/arm.md: Likewise.: Likewise.
9258 * config/arm/arm.c: Likewise.
9259 * config/arm/thumb1.md: Likewise.
9260 * config/arm/thumb2.md: Likewise.
9261 * config/arm/vfp.md: Likewise.
9262 * config/arm/arm-generic.md: Likewise.
9263 * config/arm/arm1020e.md: Likewise.
9264 * config/arm/arm1026ejs.md: Likewise.
9265 * config/arm/arm1136jfs.md: Likewise.
9266 * config/arm/arm926ejs.md: Likewise.
9267 * config/arm/cortex-a15.md: Likewise.
9268 * config/arm/cortex-a17.md: Likewise.
9269 * config/arm/cortex-a5.md: Likewise.
9270 * config/arm/cortex-a53.md: Likewise.
9271 * config/arm/cortex-a57.md: Likewise.
9272 * config/arm/cortex-a7.md: Likewise.
9273 * config/arm/cortex-a8.md: Likewise.
9274 * config/arm/cortex-a9.md: Likewise.
9275 * config/arm/cortex-m4.md: Likewise.
9276 * config/arm/cortex-m7.md: Likewise.
9277 * config/arm/cortex-r4.md: Likewise.
9278 * config/arm/exynos-m1.md: Likewise.
9279 * config/arm/fa526.md: Likewise.
9280 * config/arm/fa606te.md: Likewise.
9281 * config/arm/fa626te.md: Likewise.
9282 * config/arm/fa726te.md: Likewise.
9283 * config/arm/fmp626.md: Likewise.
9284 * config/arm/iwmmxt.md: Likewise.
9285 * config/arm/ldmstm.md: Likewise.
9286 * config/arm/marvell-pj4.md: Likewise.
9287 * config/arm/xgene1.md: Likewise.
9288 * config/aarch64/thunderx.md: Likewise.
9289 * config/aarch64/thunderx2t99.md: Likewise.
9290 * config/aarch64/falkor.md: Likewise.
9292 2017-09-12 Martin Liska <mliska@suse.cz>
9294 * attribs.c (private_lookup_attribute): New function.
9295 * attribs.h (private_lookup_attribute): Declared here.
9296 (lookup_attribute): Called from this place.
9298 2017-09-12 Richard Biener <rguenther@suse.de>
9300 PR tree-optimization/82157
9301 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9302 stmts with side-effects.
9304 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9305 Alan Hayward <alan.hayward@arm.com>
9306 David Sherwood <david.sherwood@arm.com>
9308 * target.def (hard_regno_nregs): New hook.
9309 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9310 * targhooks.h (default_hard_regno_nregs): Declare.
9311 * targhooks.c (default_hard_regno_nregs): New function.
9312 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9313 (TARGET_HARD_REGNO_NREGS): ...this hook.
9314 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9315 (CLASS_MAX_NREGS): Likewise.
9316 * doc/tm.texi: Regenerate.
9317 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9318 instead of HARD_REGNO_NREGS.
9319 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9320 HARD_REGNO_NREGS in the comment.
9321 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9322 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9323 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9324 Return an unsigned int.
9325 (TARGET_HARD_REGNO_NREGS): Redefine.
9326 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9327 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9328 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9329 (arc_hard_regno_nregs): New function.
9330 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9331 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9332 (arm_hard_regno_nregs): New function.
9333 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9334 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9335 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9336 (TARGET_HARD_REGNO_NREGS): Redefine.
9337 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9338 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9339 (HARD_REGNO_NREGS): Delete.
9340 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9341 (cr16_hard_regno_nregs): New function.
9342 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9343 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9344 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9345 (cris_hard_regno_nregs): New function.
9346 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9347 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9348 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9349 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9350 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9351 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9352 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9353 (frv_hard_regno_nregs): Make static. Take and return an
9355 (frv_class_max_nregs): Remove outdated copy of documentation.
9356 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9357 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9358 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9359 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9360 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9361 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9362 (TARGET_HARD_REGNO_NREGS): Redefine.
9363 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9364 (CLASS_MAX_NREGS): Update comment.
9365 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9366 (ia64_hard_regno_nregs): New function.
9367 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9368 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9369 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9370 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9371 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9373 (m32c_hard_regno_nregs): Likewise. Make static.
9374 (TARGET_HARD_REGNO_NREGS): Redefine.
9375 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9376 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9377 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9378 (m68k_hard_regno_nregs): New function.
9379 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9380 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9381 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9382 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9383 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9384 Take and return an unsigned int.
9385 (TARGET_HARD_REGNO_NREGS): Redefine.
9386 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9387 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9388 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9389 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9390 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9391 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9392 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9393 (msp430_hard_regno_nregs): Make static. Take and return an
9395 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9396 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9397 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9398 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9399 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9400 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9401 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9402 (TARGET_HARD_REGNO_NREGS): Redefine.
9403 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9404 (PA_HARD_REGNO_NREGS): ...this.
9405 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9406 (PA_HARD_REGNO_NREGS): ...this.
9407 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9408 (pa_hard_regno_nregs): New function.
9409 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9410 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9411 (pdp11_hard_regno_nregs): New function.
9412 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9413 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9414 (rs6000_hard_regno_nregs_hook): New function.
9415 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9416 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9417 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9418 Take and return an unsigned int. Move earlier in file.
9419 (TARGET_HARD_REGNO_NREGS): Redefine.
9420 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9421 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9422 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9423 (rl78_hard_regno_nregs): Make static. Take and return an
9425 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9426 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9427 (rs6000_hard_regno_nregs_hook): New function.
9428 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9429 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9430 (TARGET_HARD_REGNO_NREGS): Redefine.
9431 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9432 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9433 instead of HARD_REGNO_NREGS.
9434 (s390_hard_regno_nregs): New function.
9435 (s390_hard_regno_mode_ok): Add comment from s390.h.
9436 (TARGET_HARD_REGNO_NREGS): Redefine.
9437 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9438 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9439 (sh_hard_regno_nregs): New function.
9440 (sh_pass_in_reg_p): Use it.
9441 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9442 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9443 (sparc_hard_regno_nregs): New function.
9444 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9445 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9446 (spu_function_arg_advance): Use it, supplying a valid register number.
9447 (TARGET_HARD_REGNO_NREGS): Redefine.
9448 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9449 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9450 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9451 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9452 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9453 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9454 (CLASS_MAX_NREGS): Remove copy of old documentation.
9455 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9456 (visium_hard_regno_nregs): New function.
9457 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9458 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9459 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9460 xtensa_hard_regno_nregs): New function.
9461 * system.h (HARD_REGNO_NREGS): Poison.
9463 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9465 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9466 hard_regno_nregs instead of HARD_REGNO_NREGS.
9467 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9468 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9469 (c6x_expand_epilogue): Likewise.
9470 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9471 (frv_read_iacc_argument): Likewise.
9472 * config/sh/sh.c: Include regs.h.
9473 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9474 (regs_used): Likewise.
9475 (output_stack_adjust): Likewise.
9476 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9477 * expmed.c: Include regs.h.
9478 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9479 * ree.c: Include regs.h.
9480 (combine_reaching_defs): Use hard_regno_nregs instead of
9482 (add_removable_extension): Likewise.
9484 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9486 * regs.h (hard_regno_nregs): Turn into a function.
9487 (end_hard_regno): Update accordingly.
9488 * caller-save.c (setup_save_areas): Likewise.
9489 (save_call_clobbered_regs): Likewise.
9490 (replace_reg_with_saved_mem): Likewise.
9491 (insert_restore): Likewise.
9492 (insert_save): Likewise.
9493 * combine.c (can_change_dest_mode): Likewise.
9494 (move_deaths): Likewise.
9495 (distribute_notes): Likewise.
9496 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9497 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9498 (rs6000_split_multireg_move): Likewise.
9499 (rs6000_register_move_cost): Likewise.
9500 (rs6000_memory_move_cost): Likewise.
9501 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9502 (rs6000_split_multireg_move): Likewise.
9503 (rs6000_register_move_cost): Likewise.
9504 (rs6000_memory_move_cost): Likewise.
9505 * cselib.c (cselib_reset_table): Likewise.
9506 (cselib_lookup_1): Likewise.
9507 * emit-rtl.c (set_mode_and_regno): Likewise.
9508 * function.c (aggregate_value_p): Likewise.
9509 * ira-color.c (setup_profitable_hard_regs): Likewise.
9510 (check_hard_reg_p): Likewise.
9511 (calculate_saved_nregs): Likewise.
9512 (assign_hard_reg): Likewise.
9513 (improve_allocation): Likewise.
9514 (calculate_spill_cost): Likewise.
9515 * ira-emit.c (modify_move_list): Likewise.
9516 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9517 (ira_hard_reg_in_set_p): Likewise.
9518 * ira.c (setup_reg_mode_hard_regset): Likewise.
9519 (clarify_prohibited_class_mode_regs): Likewise.
9520 (check_allocation): Likewise.
9521 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9522 (lra_setup_reg_renumber): Likewise.
9523 (setup_try_hard_regno_pseudos): Likewise.
9524 (spill_for): Likewise.
9525 (assign_hard_regno): Likewise.
9526 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9527 * lra-constraints.c (in_class_p): Likewise.
9528 (lra_constraint_offset): Likewise.
9529 (simplify_operand_subreg): Likewise.
9530 (lra_constraints): Likewise.
9531 (split_reg): Likewise.
9532 (split_if_necessary): Likewise.
9533 (invariant_p): Likewise.
9534 (inherit_in_ebb): Likewise.
9535 * lra-lives.c (process_bb_lives): Likewise.
9536 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9537 (get_hard_regs): Likewise.
9538 (do_remat): Likewise.
9539 * lra-spills.c (assign_spill_hard_regs): Likewise.
9540 * mode-switching.c (create_pre_exit): Likewise.
9541 * postreload.c (reload_combine_recognize_pattern): Likewise.
9542 * recog.c (peep2_find_free_register): Likewise.
9543 * regcprop.c (kill_value_regno): Likewise.
9544 (set_value_regno): Likewise.
9545 (copy_value): Likewise.
9546 (maybe_mode_change): Likewise.
9547 (find_oldest_value_reg): Likewise.
9548 (copyprop_hardreg_forward_1): Likewise.
9549 * regrename.c (check_new_reg_p): Likewise.
9550 (regrename_do_replace): Likewise.
9551 * reload.c (push_reload): Likewise.
9552 (combine_reloads): Likewise.
9553 (find_dummy_reload): Likewise.
9554 (operands_match_p): Likewise.
9555 (find_reloads): Likewise.
9556 (find_equiv_reg): Likewise.
9557 (reload_adjust_reg_for_mode): Likewise.
9558 * reload1.c (count_pseudo): Likewise.
9559 (count_spilled_pseudo): Likewise.
9560 (find_reg): Likewise.
9561 (clear_reload_reg_in_use): Likewise.
9562 (free_for_value_p): Likewise.
9563 (allocate_reload_reg): Likewise.
9564 (choose_reload_regs): Likewise.
9565 (reload_adjust_reg_for_temp): Likewise.
9566 (emit_reload_insns): Likewise.
9567 (delete_output_reload): Likewise.
9568 * rtlanal.c (subreg_get_info): Likewise.
9569 * sched-deps.c (sched_analyze_reg): Likewise.
9570 * sel-sched.c (init_regs_for_mode): Likewise.
9571 (mark_unavailable_hard_regs): Likewise.
9572 (choose_best_reg_1): Likewise.
9573 (verify_target_availability): Likewise.
9574 * valtrack.c (dead_debug_insert_temp): Likewise.
9575 * var-tracking.c (track_loc_p): Likewise.
9576 (emit_note_insn_var_location): Likewise.
9577 * varasm.c (make_decl_rtl): Likewise.
9578 * reginfo.c (choose_hard_reg_mode): Likewise.
9579 (init_reg_modes_target): Refer directly to
9580 this_target_regs->x_hard_regno_nregs.
9582 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9584 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9585 instead of hard_regno_nregs.
9587 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9589 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9590 end_hard_regno instead of hard_regno_nregs.
9591 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9592 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9593 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9594 * ira-color.c (improve_allocation): Likewise.
9595 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9596 * lra-lives.c (mark_regno_live): Likewise.
9597 (mark_regno_dead): Likewise.
9598 * lra-remat.c (operand_to_remat): Likewise.
9599 * lra.c (collect_non_operand_hard_regs): Likewise.
9600 * postreload.c (reload_combine_note_store): Likewise.
9601 (move2add_valid_value_p): Likewise.
9602 * reload.c (regno_clobbered_p): Likewise.
9604 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9606 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9608 * config/v850/v850.c (v850_reorg): Likewise.
9609 * reload.c (refers_to_regno_for_reload_p): Likewise.
9610 (find_equiv_reg): Likewise.
9611 * reload1.c (reload_reg_reaches_end_p): Likewise.
9613 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9615 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9617 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9618 * config/arm/arm.c (output_move_neon): Likewise.
9619 (arm_attr_length_move_neon): Likewise.
9620 (neon_split_vcombine): Likewise.
9621 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9622 (c6x_mark_reg_written): Likewise.
9623 (c6x_dwarf_register_span): Likewise.
9624 * config/i386/i386.c (ix86_save_reg): Likewise.
9625 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9626 (rws_access_reg): Likewise.
9627 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9628 * mode-switching.c (create_pre_exit): Likewise.
9629 * ree.c (combine_reaching_defs): Likewise.
9630 (add_removable_extension): Likewise.
9631 * regcprop.c (find_oldest_value_reg): Likewise.
9632 (copyprop_hardreg_forward_1): Likewise.
9633 * reload.c (reload_inner_reg_of_subreg): Likewise.
9634 (push_reload): Likewise.
9635 (combine_reloads): Likewise.
9636 (find_dummy_reload): Likewise.
9637 (reload_adjust_reg_for_mode): Likewise.
9638 * reload1.c (find_reload_regs): Likewise.
9639 (forget_old_reloads_1): Likewise.
9640 (reload_reg_free_for_value_p): Likewise.
9641 (reload_adjust_reg_for_temp): Likewise.
9642 (emit_reload_insns): Likewise.
9643 (delete_output_reload): Likewise.
9644 * sel-sched.c (choose_best_reg_1): Likewise.
9645 (choose_best_pseudo_reg): Likewise.
9647 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9648 Alan Hayward <alan.hayward@arm.com>
9649 David Sherwood <david.sherwood@arm.com>
9651 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9652 * target.def (slow_unaligned_access): New hook.
9653 * targhooks.h (default_slow_unaligned_access): Declare.
9654 * targhooks.c (default_slow_unaligned_access): New function.
9655 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9656 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9657 * doc/tm.texi: Regenerate.
9658 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9659 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9660 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9662 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9663 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9665 (rs6000_slow_unaligned_access): New function.
9666 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9667 (expand_block_compare): Likewise.
9668 (expand_strn_compare): Likewise.
9669 (rs6000_rtx_costs): Likewise.
9670 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9671 (riscv_slow_unaligned_access): Likewise.
9672 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9673 (riscv_slow_unaligned_access_p): ...this and make static.
9674 (riscv_option_override): Update accordingly.
9675 (riscv_slow_unaligned_access): New function.
9676 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9677 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9678 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9679 (rs6000_slow_unaligned_access): New function.
9680 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9681 (rs6000_rtx_costs): Likewise.
9682 * config/rs6000/rs6000-string.c (expand_block_compare)
9683 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9684 of SLOW_UNALIGNED_ACCESS.
9685 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9686 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9687 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9688 of SLOW_UNALIGNED_ACCESS.
9689 * expmed.c (simple_mem_bitfield_p): Likewise.
9690 * expr.c (alignment_for_piecewise_move): Likewise.
9691 (emit_group_load_1): Likewise.
9692 (emit_group_store): Likewise.
9693 (copy_blkmode_from_reg): Likewise.
9694 (emit_push_insn): Likewise.
9695 (expand_assignment): Likewise.
9696 (store_field): Likewise.
9697 (expand_expr_real_1): Likewise.
9698 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9699 * lra-constraints.c (simplify_operand_subreg): Likewise.
9700 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9701 * gimple-ssa-store-merging.c: Likewise in block comment at start
9703 * tree-ssa-strlen.c: Include target.h.
9704 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9705 of SLOW_UNALIGNED_ACCESS.
9706 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9708 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9710 PR rtl-optimization/82185
9711 * expmed.c (emit_store_flag_int): Only test tem if it has been
9714 2017-09-12 Richard Biener <rguenther@suse.de>
9717 * match.pd ((FTYPE) N CMP CST): Fix typo.
9719 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9721 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9723 (mips_near_type_p): Add 'short_call' attribute as a synonym
9725 * doc/extend.texi (short_call): Document new function attribute.
9727 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9730 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9731 assertion check that in the condition.
9732 (get_atomic_generic_size): Likewise. Before testing if parameter
9733 has pointer type, if it has array type, call for C++
9734 default_conversion to perform array-to-pointer conversion.
9736 2017-09-12 Richard Biener <rguenther@suse.de>
9738 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9739 for operations we cannot scalarize.
9741 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9743 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9744 vectors heap vectors. Clean up comments.
9745 Make visited_bbs a reference.
9746 (profitable_jump_thread_path): Make GC
9747 vectors heap vectors. Clean up comments.
9749 (convert_and_register_jump_thread_path): Make GC vectors heap
9751 (check_subpath_and_update_thread_path): Same. Clean up comments.
9752 Make visited_bbs a reference.
9753 (handle_phi): Abstract common code to to
9754 register_jump_thread_path_if_profitable.
9755 Rename VAR_BB to DEF_BB.
9757 Make GC vectors heap vectors.
9758 Make visited_bbs a reference.
9759 (handle_assignment): Same.
9760 (register_jump_thread_path_if_profitable): New.
9761 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9763 Make GC vectors heap vectors. Clean up comments.
9764 Make visited_bbs a reference.
9765 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9766 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9769 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9772 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9773 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9775 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9777 Revert r251800 and r251799.
9779 2017-09-11 Martin Jambor <mjambor@suse.cz>
9782 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9783 arguments in advance.
9784 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9785 use it to find predecessor edges.
9786 (naive_outof_ssa): Collect vector of predecessors.
9788 2017-09-08 Jason Merrill <jason@redhat.com>
9790 PR c++/70029 - ICE with ref-qualifier and -flto
9791 * langhooks.h (struct lang_hooks_for_types): Add
9792 copy_lang_qualifiers.
9793 * attribs.c (build_type_attribute_qual_variant): Use it.
9794 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9796 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9797 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9799 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9802 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9803 (*mulsi3_sp64): New instruction.
9804 (mulsi3): New expander.
9806 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9808 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9810 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9812 * sancov.c: Include memmodel.h.
9814 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9817 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9820 2017-09-07 Carl Love <cel@us.ibm.com>
9822 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9823 the sldi instruction.
9825 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9827 * sancov.c: Include tm_p.h.
9829 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9832 * output.h (switch_to_other_text_partition): New declaration.
9833 * varasm.c (switch_to_other_text_partition): New function.
9834 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9835 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9836 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9837 to the other text partition before emitting LCL label and switch back
9838 after emitting the word after it.
9840 2017-09-07 Richard Biener <rguenther@suse.de>
9842 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9843 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9844 critical edges here, after loop init.
9845 (pass_data_pre): Remove PROP_no_crit_edges flags.
9846 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9847 for valueization of call args to avoid leaking VN_TOP.
9848 (visit_use): Assert we do not visit default defs.
9849 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9850 Use error_mark_node to more easily detect leaking VN_TOP.
9851 All default-defs are varying, not VN_TOP. Mark them visited.
9852 (run_scc_vn): Make code match comment.
9854 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9856 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9857 OPTION_MASK_FLOAT128_KEYWORD.
9858 (POWERPC_MASKS): Likewise.
9859 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9860 support for the -mfloat128-type option, and make -mfloat128
9861 default on PowerPC Linux systems. Define or undefine
9862 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9863 Define __float128 to be __ieee128 if IEEE 128-bit support is
9864 enabled, or undefine it.
9865 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9866 Delete defining __FLOAT128_TYPE__.
9867 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9868 -mfloat128-type option and make -mfloat128 default on PowerPC
9870 (TARGET_FLOAT128_TYPE): Likewise.
9871 (-mfloat128-type): Likewise.
9872 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9873 Delete the -mfloat128-type option and make -mfloat128 default on
9874 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9875 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9876 128-bit floating point is enabled. Change tests from using
9877 -mfloat128-type to -mfloat128.
9878 (rs6000_mangle_type): Use the correct mangling for the __float128
9879 type even if normal long double is restricted to 64-bits.
9880 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9882 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9883 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9884 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9886 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9887 documentation for -mfloat128.
9889 2017-09-06 Olivier Hainque <hainque@adacore.com>
9891 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9893 2017-09-06 Wish Wu <wishwu007@gmail.com>
9894 Jakub Jelinek <jakub@redhat.com>
9896 * asan.c (initialize_sanitizer_builtins): Add
9897 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9898 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9899 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9900 BT_FN_VOID_UINT64_PTR variables.
9901 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9902 (BT_FN_VOID_UINT16_UINT16): Likewise.
9903 (BT_FN_VOID_UINT32_UINT32): Likewise.
9904 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9905 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9906 (BT_FN_VOID_UINT64_PTR): Likewise.
9907 * common.opt (flag_sanitize_coverage): New variable.
9908 (fsanitize-coverage=trace-pc): Remove.
9909 (fsanitize-coverage=): Add.
9910 * flag-types.h (enum sanitize_coverage_code): New enum.
9911 * fold-const.c (fold_range_test): Disable non-short-circuit
9912 optimization if flag_sanitize_coverage.
9913 (fold_truth_andor): Likewise.
9914 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9915 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9916 (coverage_sanitizer_opts): New array.
9917 (get_closest_sanitizer_option): Add OPTS argument, handle also
9918 OPT_fsanitize_coverage_.
9919 (parse_sanitizer_options): Adjusted to also handle
9920 OPT_fsanitize_coverage_.
9921 (common_handle_option): Add OPT_fsanitize_coverage_.
9922 * sancov.c (instrument_comparison, instrument_switch): New function.
9923 (sancov_pass): Add trace-cmp support.
9924 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9925 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9926 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9927 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9928 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9929 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9930 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9931 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9932 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9933 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9935 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9937 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9938 error. Only quit immediately if parsing is complete.
9939 (BEGIN): Initialize fatal_err and parse_done.
9940 (begin fpu, end fpu): Check number of arguments.
9941 (begin arch, end arch): Likewise.
9942 (begin cpu, end cpu): Likewise.
9943 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9944 (optalias): Likewise.
9946 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9948 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9949 * config/arm/arm-isa.h: Delete. Move definitions to ...
9950 * arm-cpus.in: ... here. Use new feature and fgroup values.
9951 * config/arm/arm.c (arm_option_override): Use lower case for feature
9953 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9954 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9955 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9957 (print_isa_bits_for): New function.
9958 (gen_isa): New function.
9959 (gen_comm_data): Use print_isa_bits_for.
9960 (define feature): New keyword.
9961 (define fgroup): New keyword.
9962 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9963 (arm-isa.h): Add rule to generate file.
9964 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9965 case for feature bit names.
9967 2017-09-06 Richard Biener <rguenther@suse.de>
9969 * tree-ssa-pre.c (NECESSARY): Remove.
9970 (create_expression_by_pieces): Do not touch pass-local flags.
9971 (insert_into_preds_of_block): Likewise.
9972 (do_pre_regular_insertion): Likewise.
9973 (eliminate_insert): Likewise.
9974 (eliminate_dom_walker::before_dom_children): Likewise.
9975 (fini_eliminate): Do not look at inserted_exprs.
9976 (mark_operand_necessary): Remove.
9977 (remove_dead_inserted_code): Replace with simple work-list
9978 algorithm based on inserted_exprs and SSA uses.
9979 (pass_pre::execute): Re-order fini_eliminate and
9980 remove_dead_inserted_code.
9982 2017-09-06 Olivier Hainque <hainque@adacore.com>
9984 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9985 for VxWorks 7. Adjust surrounding comments.
9987 2017-09-06 Richard Biener <rguenther@suse.de>
9989 * gimple-ssa-strength-reduction.c
9990 (find_candidates_dom_walker::before_dom_children): Also allow
9993 2017-09-06 Richard Biener <rguenther@suse.de>
9995 PR tree-optimization/82108
9996 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9997 for gap in the non-permutation SLP case.
9999 2017-09-06 Martin Jambor <mjambor@suse.cz>
10001 PR tree-optimization/82078
10002 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10003 add_access_to_work_queue...
10004 (build_accesses_from_assign): ...here.
10005 (propagate_all_subaccesses): Make sure racc is the group
10006 representative, if there is one.
10008 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10010 PR middle-end/82095
10011 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10014 2017-09-06 Richard Biener <rguenther@suse.de>
10016 * gimple-ssa-strength-reduction.c
10017 (find_candidates_dom_walker::before_doom_children): Use a
10018 type and not a mode check.
10020 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10023 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10025 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10027 2017-09-05 Jeff Law <law@redhat.com>
10029 PR tree-optimization/64910
10030 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10031 cases where we have 3 or more operands.
10033 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10035 PR middle-end/81768
10036 * omp-low.c (lower_omp_for): Recompute tree invariant if
10037 gimple_omp_for_initial/final is ADDR_EXPR.
10039 PR middle-end/81768
10040 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10041 into gimple val before gimplification fo the COND_EXPR.
10043 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10045 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10046 REGION_COPY argument.
10047 (thread_through_all_blocks): Remove unused argument to
10048 duplicate_thread_path.
10050 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10051 Alan Hayward <alan.hayward@arm.com>
10052 David Sherwood <david.sherwood@arm.com>
10054 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10055 Take a scalar_mode rather than a machine_mode.
10056 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10057 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10058 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10059 (aarch64_gen_adjusted_ldpstp): Likewise.
10060 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10062 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10063 Alan Hayward <alan.hayward@arm.com>
10064 David Sherwood <david.sherwood@arm.com>
10066 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10067 Take a scalar_int_mode instead of a machine_mode.
10068 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10069 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10070 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10071 (aarch64_simd_attr_length_rglist): Delete.
10072 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10073 a scalar_int_mode instead of a machine_mode.
10074 (aarch64_add_offset): Likewise.
10075 (aarch64_internal_mov_immediate): Likewise
10076 (aarch64_add_constant_internal): Likewise.
10077 (aarch64_add_constant): Likewise.
10078 (aarch64_movw_imm): Likewise.
10079 (aarch64_rtx_arith_op_extract_p): Likewise.
10080 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10081 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10082 Remove assert that the mode isn't a vector.
10083 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10084 (aarch64_expand_mov_immediate): Update calls after above changes.
10085 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10086 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10087 (aarch64_move_imm): Likewise.
10088 (aarch64_can_const_movi_rtx_p): Likewise.
10089 (aarch64_strip_extend): Likewise.
10090 (aarch64_extr_rtx_p): Likewise.
10091 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10092 a CONST_INT when the mode parameter is VOIDmode.
10093 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10095 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10097 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10098 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10099 (bitwise_type_for_mode): Update accordingly.
10101 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10103 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10104 * stor-layout.c (mode_for_size_tree): Likewise.
10105 (mode_for_array): Update accordingly.
10106 (layout_decl): Likewise.
10107 (compute_record_mode): Likewise. Only set the mode once.
10109 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10111 * target.def (get_mask_mode): Change return type to opt_mode.
10113 * doc/tm.texi: Regenerate.
10114 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10115 * targhooks.c (default_get_mask_mode): Likewise.
10116 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10117 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10118 targetm.get_mask_mode.
10119 * tree.c (build_truth_vector_type): Likewise.
10121 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10123 * machmode.h (mode_for_vector): Return an opt_mode.
10124 * stor-layout.c (mode_for_vector): Likewise.
10125 (mode_for_int_vector): Update accordingly.
10126 (layout_type): Likewise.
10127 * config/i386/i386.c (emit_memmov): Likewise.
10128 (ix86_expand_set_or_movmem): Likewise.
10129 (ix86_expand_vector_init): Likewise.
10130 (ix86_get_mask_mode): Likewise.
10131 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10133 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10134 * expmed.c (extract_bit_field_1): Likewise.
10135 * expr.c (expand_expr_real_2): Likewise.
10136 * optabs-query.c (can_vec_perm_p): Likewise.
10137 (can_vec_mask_load_store_p): Likewise.
10138 * optabs.c (expand_vec_perm): Likewise.
10139 * targhooks.c (default_get_mask_mode): Likewise.
10140 * tree-vect-stmts.c (vectorizable_store): Likewise.
10141 (vectorizable_load): Likewise.
10142 (get_vectype_for_scalar_type_and_size): Likewise.
10144 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10146 * machmode.h (mode_for_int_vector): New function.
10147 * stor-layout.c (mode_for_int_vector): Likewise.
10148 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10149 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10150 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10151 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10152 (s390_expand_vcond): Likewise.
10154 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10156 * machmode.h (opt_machine_mode): New type.
10157 (opt_mode<T>): Allow construction from anything that can be
10159 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10160 (mode_for_size): Return an opt_machine_mode.
10161 * stor-layout.c (mode_for_size): Likewise.
10162 (mode_for_size_tree): Update call accordingly.
10163 (bitwise_mode_for_mode): Likewise.
10164 (make_fract_type): Likewise.
10165 (make_accum_type): Likewise.
10166 * caller-save.c (replace_reg_with_saved_mem): Update call
10168 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10169 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10170 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10171 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10172 * expmed.c (extract_bit_field_1): Likewise.
10173 * reload.c (get_secondary_mem): Likewise.
10174 * varasm.c (assemble_integer): Likewise.
10175 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10178 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10180 * machmode.h (decimal_float_mode_for_size): New function.
10181 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10182 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10183 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10184 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10185 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10186 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10188 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10190 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10191 (get_builtin_sync_mode): Likewise.
10192 (expand_ifn_atomic_compare_exchange): Likewise.
10193 (expand_builtin_atomic_clear): Likewise.
10194 (expand_builtin_atomic_test_and_set): Likewise.
10195 (fold_builtin_atomic_always_lock_free): Likewise.
10196 * calls.c (compute_argument_addresses): Likewise.
10197 (emit_library_call_value_1): Likewise.
10198 (store_one_arg): Likewise.
10199 * combine.c (combine_instructions): Likewise.
10200 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10201 * config/arm/arm.c (arm_function_value): Likewise.
10202 (aapcs_allocate_return_reg): Likewise.
10203 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10204 * config/i386/i386.c (construct_container): Likewise.
10205 (ix86_gimplify_va_arg): Likewise.
10206 (ix86_expand_sse_cmp): Likewise.
10207 (emit_memmov): Likewise.
10208 (emit_memset): Likewise.
10209 (expand_small_movmem_or_setmem): Likewise.
10210 (ix86_expand_pextr): Likewise.
10211 (ix86_expand_pinsr): Likewise.
10212 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10213 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10215 * config/mips/mips.c (mips_function_value_1) Likewise.
10216 (mips_block_move_straight): Likewise.
10217 (mips_expand_ins_as_unaligned_store): Likewise.
10218 * config/powerpcspe/powerpcspe.c
10219 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10220 (rs6000_darwin64_record_arg_flush): Likewise.
10221 * config/rs6000/rs6000.c
10222 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10223 (rs6000_darwin64_record_arg_flush): Likewise.
10224 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10225 (sparc_function_value_1): Likewise.
10226 * config/spu/spu.c (adjust_operand): Likewise.
10227 (spu_emit_branch_or_set): Likewise.
10228 (arith_immediate_p): Likewise.
10229 * emit-rtl.c (gen_lowpart_common): Likewise.
10230 * expr.c (expand_expr_real_1): Likewise.
10231 * function.c (assign_parm_setup_block): Likewise.
10232 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10233 * reload1.c (alter_reg): Likewise.
10234 * stor-layout.c (mode_for_vector): Likewise.
10235 (layout_type): Likewise.
10237 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10239 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10240 (spu_convert_move): Likewise.
10241 * lower-subreg.c (resolve_simple_move): Likewise.
10243 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10246 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10247 define_insn to a define_expand.
10248 (altivec_vsum2sws_direct): New define_insn.
10249 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10251 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10253 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10256 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10261 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10262 realign stack if stack alignment needed is less than incoming
10265 2017-09-05 Marek Polacek <polacek@redhat.com>
10268 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10271 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10273 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10275 2017-09-05 Richard Biener <rguenther@suse.de>
10277 PR tree-optimization/82084
10278 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10280 2017-09-05 Richard Biener <rguenther@suse.de>
10282 PR tree-optimization/82102
10283 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10285 2017-09-05 Martin Liska <mliska@suse.cz>
10287 PR tree-optimization/82032
10288 * tree-cfg.c (generate_range_test): New function.
10289 * tree-cfg.h (generate_range_test): Declared here.
10290 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10291 (cleanup_control_expr_graph): Use it.
10292 * tree-switch-conversion.c (try_switch_expansion): Remove
10294 (emit_case_nodes): Use generate_range_test.
10296 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10299 * config/i386/i386.md (*<btsc><mode>_mask): Add
10300 TARGET_USE_BT to insn constraint.
10301 (*btr<mode>_mask): Ditto.
10303 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10305 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10306 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10308 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10311 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10312 TARGET_NEON and TARGET_IWMMXT.
10313 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10314 TARGET_NEON and TARGET_IWMMXT.
10315 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10317 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10319 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10320 (ix86_rewrite_tls_address): Ditto.
10321 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10322 (ix86_rewrite_tls_address_1): Ditto.
10323 (ix86_rewrite_tls_address): Ditto.
10324 * config/i386/predicates.md (tls_address_pattern): New predicate.
10325 * config/i386/i386.md (TLS address splitter): New splitter.
10327 2017-09-04 Richard Biener <rguenther@suse.de>
10329 PR tree-optimization/82084
10330 * fold-const.h (can_native_encode_string_p): Declare.
10331 * fold-const.c (can_native_encode_string_p): Factor out from ...
10332 (native_encode_string): ... here.
10333 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10334 vectorizing stores from constants we later cannot handle.
10336 2017-09-04 Marek Polacek <polacek@redhat.com>
10339 * doc/invoke.texi: Update -Wtautological-compare documentation.
10341 2017-09-04 Jeff Law <law@redhat.com>
10343 PR tree-optimization/64910
10344 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10345 swap the first and last operand if the last is a constant.
10347 2017-09-04 Marek Polacek <polacek@redhat.com>
10350 * convert.c (do_narrow): When sanitizing signed integer overflows,
10351 bail out for signed types.
10352 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10354 2017-09-04 Richard Biener <rguenther@suse.de>
10356 PR tree-optimization/82060
10357 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10358 Move devirtualization after stmt folding and before EH/AB/noreturn
10359 cleanup to get the stmt refs canonicalized. Use a bool instead
10360 of gimple_modified_p since that doesn't work for NOPs. Schedule
10361 NOPs generated by folding for removal.
10363 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10364 Alan Hayward <alan.hayward@arm.com>
10365 David Sherwood <david.sherwood@arm.com>
10367 * coretypes.h (pad_direction): New enum.
10368 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10369 (FUNCTION_ARG_PADDING): Likewise.
10370 * target.def (function_arg_padding): New hook.
10371 * targhooks.h (default_function_arg_padding): Declare.
10372 * targhooks.c (default_function_arg_padding): New function.
10373 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10374 (TARGET_FUNCTION_ARG_PADDING): ...this.
10375 * doc/tm.texi: Regenerate.
10376 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10377 instead of direction.
10378 (compute_argument_addresses): Likewise.
10379 (load_register_parameters): Likewise.
10380 (emit_library_call_value_1): Likewise.
10381 (store_one_arg): Use targetm.calls.function_arg_padding instead
10382 of FUNCTION_ARG_PADDING.
10383 (must_pass_in_stack_var_size_or_pad): Likewise.
10384 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10385 (emit_group_store): Likewise.
10386 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10387 instead of FUNCTION_ARG_PADDING.
10388 (emit_push_insn): Likewise, and propagate enum change throughout
10390 * function.h (direction): Delete.
10391 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10393 * function.c (assign_parm_find_stack_rtl): Likewise.
10394 (assign_parm_setup_block_p): Likewise.
10395 (assign_parm_setup_block): Likewise.
10396 (gimplify_parameters): Likewise.
10397 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10398 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10400 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10401 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10402 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10403 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10404 (aarch64_function_arg_padding): ...this new function.
10405 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10406 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10407 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10408 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10409 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10410 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10411 (arm_pad_arg_upward): Replace with...
10412 (arm_function_arg_padding): ...this new function.
10413 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10415 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10416 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10417 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10418 (ia64_hpux_function_arg_padding): Replace with...
10419 (ia64_function_arg_padding): ...this new function. Use pad_direction
10420 instead of direction. Check for TARGET_HPUX.
10421 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10422 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10423 (iq2000_function_arg_padding): New function.
10424 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10425 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10426 (mips_function_arg_padding): ...this new function.
10427 (mips_pad_reg_upward): Update accordingly.
10428 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10429 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10430 targetm.calls.function_arg_padding.
10431 (FUNCTION_ARG_PADDING): Delete.
10432 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10433 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10434 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10435 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10436 (nios2_block_reg_padding): Return pad_direction instead of direction.
10437 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10438 instead of direction.
10439 (nios2_function_arg_padding): Likewise. Make static.
10440 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10441 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10442 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10443 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10444 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10445 (pa_function_arg_padding): Make static. Return pad_direction instead
10447 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10448 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10449 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10450 instead of direction. Use targetm.calls.function_arg_padding.
10451 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10452 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10453 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10454 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10455 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10457 (function_arg_padding): Rename to...
10458 (rs6000_function_arg_padding): ...this. Make static. Return
10459 pad_direction instead of direction.
10460 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10461 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10462 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10463 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10464 instead of direction. Use targetm.calls.function_arg_padding.
10465 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10466 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10467 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10468 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10469 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10470 (function_arg_padding): Rename to...
10471 (rs6000_function_arg_padding): ...this. Make static. Return
10472 pad_direction instead of direction.
10473 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10474 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10475 * config/s390/s390.c (s390_function_arg_padding): New function.
10476 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10477 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10478 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10479 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10480 (function_arg_padding): Rename to...
10481 (sparc_function_arg_padding): ...this. Make static. Return
10482 pad_direction instead of direction.
10483 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10484 * config/spu/spu.c (spu_function_arg_padding): New function.
10485 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10486 * system.h (FUNCTION_ARG_PADDING): Poison.
10488 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10489 Alan Hayward <alan.hayward@arm.com>
10490 David Sherwood <david.sherwood@arm.com>
10492 * target.def (modes_tieable_p): New hook.
10493 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10494 (TARGET_MODES_TIEABLE_P): ...this.
10495 * doc/tm.texi.in: Regenerate.
10496 * hooks.h (hook_bool_mode_mode_true): Declare.
10497 * hooks.c (hook_bool_mode_mode_true): New function.
10498 * combine.c (subst): Use targetm.modes_tieable_p instead of
10500 * dse.c (find_shift_sequence): Likewise.
10501 * expmed.c (extract_low_bits): Likewise.
10502 * lower-subreg.c: Include target.h.
10503 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10505 * rtlanal.c (rtx_cost): Likewise.
10506 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10507 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10508 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10509 (TARGET_MODES_TIEABLE_P): Redefine.
10510 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10511 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10512 (TARGET_MODES_TIEABLE_P): Redefine.
10513 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10514 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10515 (arc_modes_tieable_p): New function.
10516 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10517 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10518 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10519 (arm_modes_tieable_p): Make static.
10520 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10521 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10522 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10523 (TARGET_MODES_TIEABLE_P): Redefine.
10524 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10525 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10526 (TARGET_MODES_TIEABLE_P): Redefine.
10527 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10528 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10529 (cr16_modes_tieable_p): New function.
10530 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10531 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10532 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10533 (TRULY_NOOP_TRUNCATION): Update comment.
10534 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10535 (TRULY_NOOP_TRUNCATION): Update comment.
10536 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10537 (frv_modes_tieable_p): New function.
10538 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10539 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10540 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10541 (TARGET_MODES_TIEABLE_P): Redefine.
10542 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10543 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10544 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10545 (TARGET_MODES_TIEABLE_P): Redefine.
10546 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10547 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10548 (ia64_modes_tieable_p): New function.
10549 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10550 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10551 (iq2000_modes_tieable_p): New function.
10552 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10553 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10554 (lm32_modes_tieable_p): New function.
10555 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10556 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10557 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10558 (TARGET_MODES_TIEABLE_P): Redefine.
10559 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10560 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10561 (m32r_modes_tieable_p): New function.
10562 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10563 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10564 (m68k_modes_tieable_p): New function.
10565 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10566 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10567 (mcore_modes_tieable_p): New function.
10568 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10569 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10571 (TARGET_MODES_TIEABLE_P): Redefine.
10572 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10573 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10574 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10575 (TARGET_MODES_TIEABLE_P): Redefine.
10576 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10577 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10578 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10579 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10580 (mn10300_modes_tieable_p): ...this and make static.
10581 (TARGET_MODES_TIEABLE_P): Redefine.
10582 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10583 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10584 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10585 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10586 (msp430_modes_tieable_p): Make static.
10587 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10588 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10589 (TARGET_MODES_TIEABLE_P): Redefine.
10590 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10591 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10592 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10593 (TARGET_MODES_TIEABLE_P): Redefine.
10594 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10595 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10596 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10597 (TARGET_MODES_TIEABLE_P): Redefine.
10598 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10599 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10600 (pdp11_modes_tieable_p): New function.
10601 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10602 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10603 (rs6000_modes_tieable_p): New function.
10604 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10605 * config/powerpcspe/powerpcspe.md: Update comment.
10606 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10607 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10608 (TARGET_MODES_TIEABLE_P): Redefine.
10609 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10610 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10611 (rl78_modes_tieable_p): New function.
10612 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10613 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10614 (rs6000_modes_tieable_p): New function.
10615 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10616 * config/rs6000/rs6000.md: Update comment.
10617 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10618 * config/rx/rx.c (rx_modes_tieable_p): New function.
10619 (TARGET_MODES_TIEABLE_P): Redefine.
10620 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10621 * config/s390/s390.c (s390_modes_tieable_p): New function.
10622 (TARGET_MODES_TIEABLE_P): Redefine.
10623 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10624 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10625 (sh_modes_tieable_p): New function.
10626 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10627 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10628 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10629 (sparc_modes_tieable_p): Make static.
10630 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10631 * config/spu/spu.c (spu_modes_tieable_p): New function.
10632 (TARGET_MODES_TIEABLE_P): Redefine.
10633 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10634 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10635 (TARGET_MODES_TIEABLE_P): Redefine.
10636 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10637 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10638 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10639 * config/v850/v850.c (v850_modes_tieable_p): New function.
10640 (TARGET_MODES_TIEABLE_P): Redefine.
10641 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10642 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10643 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10644 (visium_modes_tieable_p): New function.
10645 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10646 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10647 (xtensa_modes_tieable_p): New function.
10648 * system.h (MODES_TIEABLE_P): Poison.
10650 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10651 Alan Hayward <alan.hayward@arm.com>
10652 David Sherwood <david.sherwood@arm.com>
10654 * target.def (hard_regno_mode_ok): New hook.
10655 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10656 (TARGET_HARD_REGNO_MODE_OK): ...this.
10657 * doc/tm.texi.in: Regenerate.
10658 * hooks.h (hook_bool_uint_mode_true): Declare.
10659 * hooks.c (hook_bool_uint_mode_true): New function.
10660 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10661 HARD_REGNO_MODE_OK.
10662 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10663 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10664 instead of HARD_REGNO_MODE_OK.
10665 * caller-save.c: Include target.h.
10666 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10667 HARD_REGNO_MODE_OK.
10668 * combine.c (can_combine_p): Likewise.
10669 (combinable_i3pat): Likewise.
10670 (can_change_dest_mode): Likewise.
10671 * expr.c (init_expr_target): Likewise.
10672 (convert_move): Likewise.
10673 (convert_modes): Likewise.
10674 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10675 (setup_prohibited_mode_move_regs): Likewise.
10676 * ira.h (target_ira): Likewise.
10677 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10678 * lra-constraints.c (process_alt_operands): Likewise.
10679 (split_reg): Likewise.
10680 * recog.c (peep2_find_free_register): Likewise.
10681 * ree.c (combine_reaching_defs): Likewise.
10682 * regcprop.c (maybe_mode_change): Likewise.
10683 * reginfo.c (init_reg_sets_1): Likewise.
10684 (choose_hard_reg_mode): Likewise.
10685 (simplifiable_subregs): Likewise.
10686 * regrename.c (check_new_reg_p): Likewise.
10687 * reload.c (find_valid_class): Likewise.
10688 (find_valid_class_1): Likewise.
10689 (reload_inner_reg_of_subreg): Likewise.
10690 (push_reload): Likewise.
10691 (combine_reloads): Likewise.
10692 (find_dummy_reload): Likewise.
10693 (find_reloads): Likewise.
10694 * reload1.c (find_reg): Likewise.
10695 (set_reload_reg): Likewise.
10696 (allocate_reload_reg): Likewise.
10697 (choose_reload_regs): Likewise.
10698 (reload_adjust_reg_for_temp): Likewise.
10699 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10700 (simplify_subreg_regno): Likewise.
10701 * sel-sched.c (init_regs_for_mode): Likewise.
10702 * varasm.c (make_decl_rtl): Likewise.
10703 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10704 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10705 HARD_REGNO_MODE_OK.
10706 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10707 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10708 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10709 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10710 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10711 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10712 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10713 (arc_mode_class): Delete.
10714 (HARD_REGNO_MODE_OK): Delete.
10715 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10716 (arc_hard_regno_mode_ok): Rename old array to...
10717 (arc_hard_regno_mode_ok_modes): ...this.
10718 (arc_conditional_register_usage): Update accordingly.
10719 (arc_mode_class): Make static.
10720 (arc_hard_regno_mode_ok): New function.
10721 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10722 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10723 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10724 (arm_hard_regno_mode_ok): Make static.
10725 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10726 HARD_REGNO_MODE_OK.
10727 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10728 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10729 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10731 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10732 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10733 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10734 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10735 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10737 * config/bfin/predicates.md (valid_reg_operand): Use
10738 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10739 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10740 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10741 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10742 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10743 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10744 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10745 (cr16_hard_regno_mode_ok): Make static and return a bool.
10746 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10747 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10748 (cris_hard_regno_mode_ok): New function.
10749 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10750 (epiphany_mode_class): Delete.
10751 (HARD_REGNO_MODE_OK): Delete.
10752 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10753 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10754 (hard_regno_mode_ok): Rename to...
10755 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10756 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10757 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10758 HARD_REGNO_MODE_OK.
10759 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10760 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10761 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10762 (frv_hard_regno_mode_ok): Make static and return a bool.
10763 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10764 HARD_REGNO_MODE_OK.
10765 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10766 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10767 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10768 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10770 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10771 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10772 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10773 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10776 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10777 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10778 (ia64_hard_regno_mode_ok): New function.
10779 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10780 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10781 (iq2000_hard_regno_mode_ok): New function.
10782 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10783 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10784 (lm32_hard_regno_mode_ok): New function.
10785 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10786 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10787 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10788 instead of HARD_REGNO_MODE_OK.
10789 (m32c_hard_regno_ok): Rename to...
10790 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10791 (m32c_cannot_change_mode_class): Update accordingly.
10792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10793 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10794 (m32r_mode_class): Delete.
10795 (HARD_REGNO_MODE_OK): Delete.
10796 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10797 (m32r_hard_regno_mode_ok): Rename to...
10798 (m32r_hard_regno_modes): ...this.
10799 (m32r_mode_class): Make static.
10800 (m32r_hard_regno_mode_ok): New function.
10801 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10802 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10803 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10804 (m68k_hard_regno_mode_ok): Make static.
10805 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10806 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10807 (mcore_hard_regno_mode_ok): New function.
10808 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10809 (HARD_REGNO_MODE_OK): Delete.
10810 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10812 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10813 (microblaze_hard_regno_mode_ok): New function.
10814 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10815 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10816 (mips_hard_regno_mode_ok): Delete.
10817 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10818 (mips_hard_regno_mode_ok_p): ...this and make static.
10819 (mips_hard_regno_mode_ok_p): Rename to...
10820 (mips_hard_regno_mode_ok_uncached): ...this.
10821 (mips_hard_regno_mode_ok): New function.
10822 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10823 of HARD_REGNO_MODE_OK.
10824 (mips_option_override): Update after above name changes.
10825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10826 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10827 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10828 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10829 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10831 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10832 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10833 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10834 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10835 (msp430_hard_regno_mode_ok): Make static and return a bool.
10836 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10837 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10838 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10840 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10841 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10842 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10843 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10844 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10845 (PA_HARD_REGNO_MODE_OK): ...this
10846 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10847 (PA_HARD_REGNO_MODE_OK): ...this.
10848 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10849 (pa_hard_regno_mode_ok): New function.
10850 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10851 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10852 (pdp11_hard_regno_mode_ok): New function.
10853 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10854 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10856 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10858 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10859 (rs6000_hard_regno_mode_ok): Rename to...
10860 (rs6000_hard_regno_mode_ok_uncached): ...this.
10861 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10862 (rs6000_hard_regno_mode_ok): New function.
10863 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10864 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10865 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10866 (riscv_hard_regno_mode_ok): ...this and make static.
10867 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10868 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10869 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10870 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10871 (rl78_hard_regno_mode_ok): Make static and return bool.
10872 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10873 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10875 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10876 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10877 (rs6000_hard_regno_mode_ok): Rename to...
10878 (rs6000_hard_regno_mode_ok_uncached): ...this.
10879 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10880 (rs6000_hard_regno_mode_ok): New function.
10881 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10882 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10884 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10885 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10886 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10887 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10888 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10889 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10890 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10891 (sh_hard_regno_mode_ok): Make static.
10892 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10893 instead of HARD_REGNO_MODE_OK.
10894 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10895 (sparc_mode_class): Delete.
10896 (HARD_REGNO_MODE_OK): Delete.
10897 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10898 (hard_regno_mode_classes): Make static.
10899 (sparc_mode_class): Likewise.
10900 (sparc_hard_regno_mode_ok): New function.
10901 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10902 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10903 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10905 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10906 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10907 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10908 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10909 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10911 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10912 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10913 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10914 (visium_hard_regno_mode_ok): New function.
10915 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10916 instead of HARD_REGNO_MODE_OK.
10917 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10918 (HARD_REGNO_MODE_OK): Delete.
10919 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10920 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10921 (xtensa_option_override): Update accordingly.
10922 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10923 (xtensa_hard_regno_mode_ok): New function.
10924 * system.h (HARD_REGNO_MODE_OK): Poison.
10926 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10927 Alan Hayward <alan.hayward@arm.com>
10928 David Sherwood <david.sherwood@arm.com>
10930 * target.def (hard_regno_call_part_clobbered): New hook.
10931 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10932 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10933 * doc/tm.texi: Regenerate.
10934 * hooks.h (hook_bool_uint_mode_false): Declare.
10935 * hooks.c (hook_bool_uint_mode_false): New function.
10936 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10937 * cselib.c (cselib_process_insn): Use
10938 targetm.hard_regno_call_part_clobbered instead of
10939 HARD_REGNO_CALL_PART_CLOBBERED.
10940 * ira-conflicts.c (ira_build_conflicts): Likewise.
10941 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10942 * lra-constraints.c (need_for_call_save_p): Likewise.
10943 * lra-lives.c: Include target.h.
10944 (check_pseudos_live_through_calls): Use
10945 targetm.hard_regno_call_part_clobbered instead of
10946 HARD_REGNO_CALL_PART_CLOBBERED.
10947 * regcprop.c: Include target.h.
10948 (copyprop_hardreg_forward_1): Use
10949 targetm.hard_regno_call_part_clobbered instead of
10950 HARD_REGNO_CALL_PART_CLOBBERED.
10951 * reginfo.c (choose_hard_reg_mode): Likewise.
10952 * regrename.c (check_new_reg_p): Likewise.
10953 * reload.c (find_equiv_reg): Likewise.
10954 * reload1.c (emit_reload_insns): Likewise.
10955 * sched-deps.c (deps_analyze_insn): Likewise.
10956 * sel-sched.c (init_regs_for_mode): Likewise.
10957 (mark_unavailable_hard_regs): Likewise.
10958 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10959 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10960 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10962 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10963 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10964 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10966 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10968 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10969 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10970 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10972 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10973 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10974 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10976 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10977 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10979 * config/powerpcspe/powerpcspe.c
10980 (rs6000_hard_regno_call_part_clobbered): New function.
10981 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10982 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10983 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10985 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10986 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10987 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10989 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10990 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10991 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10993 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * rtl.h (subreg_memory_offset): Declare.
10998 * emit-rtl.c (subreg_memory_offset): New function.
10999 * expmed.c (store_bit_field_1): Use it.
11000 * expr.c (undefined_operand_subword_p): Likewise.
11001 * simplify-rtx.c (simplify_subreg): Likewise.
11003 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11005 PR rtl-optimization/57448
11008 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11009 using atomic_load pattern.
11010 (expand_atomic_store): Likewise.
11012 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11015 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11016 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11017 handling. Use replace_call_with_value with NULL instead of
11018 gsi_replace, unlink_stmt_vdef and release_defs.
11020 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11023 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11025 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11028 * rtl.h (emit_library_call_value_1): Declare.
11029 (emit_library_call): Replace declaration with a series of overloads.
11030 Remove the parameter count argument.
11031 (emit_library_call_value): Likewise.
11032 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11033 with an "rtx_mode_t *".
11034 (emit_library_call_value): Delete.
11035 (emit_library_call): Likewise.
11036 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11037 (asan_emit_allocas_unpoison): Likewise.
11038 * builtins.c (expand_builtin_powi): Likewise.
11039 (expand_asan_emit_allocas_unpoison): Likewise.
11040 * cfgexpand.c (expand_main_function): Likewise.
11041 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11042 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11043 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11044 * config/arm/arm.c (arm_trampoline_init): Likewise.
11045 (arm_call_tls_get_addr): Likewise.
11046 (arm_expand_divmod_libfunc): Likewise.
11047 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11048 (smulsi3_highpart): Likewise.
11049 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11050 (c6x_expand_compare): Likewise.
11051 (c6x_expand_movmem): Likewise.
11052 * config/frv/frv.c (frv_trampoline_init): Likewise.
11053 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11054 (ix86_expand_divmod_libfunc): Likewise.
11055 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11056 (ia64_expand_compare): Likewise.
11057 (ia64_profile_hook): Likewise.
11058 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11059 (nonlocal_goto): Likewise.
11060 (restore_stack_nonlocal): Likewise.
11061 * config/m32r/m32r.c (block_move_call): Likewise.
11062 (m32r_trampoline_init): Likewise.
11063 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11064 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11065 (m68k_call_m68k_read_tp): Likewise.
11066 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11067 (microblaze_expand_divide): Likewise.
11068 * config/mips/mips.h (mips_args): Likewise.
11069 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11070 (MIPS_ICACHE_SYNC): Likewise.
11071 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11072 (nios2_trampoline_init): Likewise.
11073 * config/pa/pa.c (hppa_tls_call): Likewise.
11074 (pa_trampoline_init): Likewise.
11075 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11076 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11077 (expand_strn_compare): Likewise.
11078 (rs6000_generate_compare): Likewise.
11079 (rs6000_expand_float128_convert): Likewise.
11080 (output_profile_hook): Likewise.
11081 (rs6000_trampoline_init): Likewise.
11082 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11083 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11084 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11085 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11086 (rs6000_generate_compare): Likewise.
11087 (rs6000_expand_float128_convert): Likewise.
11088 (output_profile_hook): Likewise.
11089 (rs6000_trampoline_init): Likewise.
11090 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11091 * config/sh/sh.c (sh_trampoline_init): Likewise.
11092 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11093 (sparc_emit_float_lib_cmp): Likewise.
11094 (sparc32_initialize_trampoline): Likewise.
11095 (sparc64_initialize_trampoline): Likewise.
11096 (sparc_profile_hook): Likewise.
11097 * config/spu/spu.c (ea_load_store): Likewise.
11098 * config/spu/spu.md (floatunssidf2): Likewise.
11099 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11100 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11101 * config/visium/visium.c (expand_block_move_4): Likewise.
11102 (expand_block_move_2): Likewise.
11103 (expand_block_move_1): Likewise.
11104 (expand_block_set_4): Likewise.
11105 (expand_block_set_2): Likewise.
11106 (expand_block_set_1): Likewise.
11107 (visium_trampoline_init): Likewise.
11108 (visium_profile_hook): Likewise.
11109 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11110 (xtensa_setup_frame_addresses): Likewise.
11111 (xtensa_trampoline_init): Likewise.
11112 * except.c (sjlj_emit_function_enter): Likewise.
11113 (sjlj_emit_function_exit): Likewise.
11114 * explow.c (allocate_dynamic_stack_space): Likewise.
11115 (probe_stack_range): Likewise.
11116 * expr.c (convert_mode_scalar): Likewise.
11117 * optabs.c (expand_binop): Likewise.
11118 (expand_twoval_binop_libfunc): Likewise.
11119 (expand_unop): Likewise.
11120 (prepare_cmp_insn): Likewise.
11121 (prepare_float_lib_cmp): Likewise.
11122 (expand_float): Likewise.
11123 (expand_fix): Likewise.
11124 (expand_fixed_convert): Likewise.
11125 (maybe_emit_sync_lock_test_and_set): Likewise.
11126 (expand_atomic_compare_and_swap): Likewise.
11127 (expand_mem_thread_fence): Likewise.
11128 (expand_atomic_fetch_op): Likewise.
11130 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11132 * doc/generic.texi (OpenACC): Adjust URL.
11133 * doc/invoke.texi (C Dialect Options): Ditto.
11135 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11137 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11138 predicate for operand 1. Add (m,<S>) constraint.
11139 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11140 Prevent memory operand 1 with register operand 2.
11142 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11144 PR rtl-optimization/82024
11145 * combine.c (try_combine): If the combination result is a PARALLEL,
11146 and we only need to retain the SET in there that would be placed
11147 at I2, check that we can place that at I3 instead, before doing so.
11149 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11152 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11154 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11155 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11158 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11159 Jeff Law <law@redhat.com>
11161 * varasm.c (bss_initializer_p): Do not put constants into .bss
11162 (categorize_decl_for_section): Handle bss_initializer_p returning
11163 false when DECL_INITIAL is NULL.
11165 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11168 * config/s390/s390.c (s390_can_inline_p): New function.
11170 2017-09-01 Jeff Law <law@redhat.com>
11172 PR tree-optimization/82052
11173 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11174 Always initialize the returned slot after a hash table miss
11175 when INSERT is true.
11177 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11179 * config/s390/s390.md (mem_signal_fence): Remove.
11180 * doc/md.texi (mem_signal_fence): Remove.
11181 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11183 * target-insns.def (mem_signal_fence): Remove.
11185 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11188 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11191 * asan.c (create_odr_indicator): Strip name encoding from assembler
11192 name before appending it after __odr_asan_.
11194 2017-09-01 Martin Liska <mliska@suse.cz>
11196 PR tree-optimization/82059
11197 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11198 frequency only when an edge is redirected.
11200 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11202 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11203 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11204 (arc_conditional_register_usage): Remove ARC600 lp_count
11206 (arc_file_start): Emit Tag_ARC_CPU_variation.
11207 (arc_can_use_doloop_p): New conditions to use ZOLs.
11208 (hwloop_fail): New function.
11209 (hwloop_optimize): Likewise.
11210 (hwloop_pattern_reg): Likewise.
11211 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11212 (arc_reorg_loops): New function, calls reorg_loops.
11213 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11214 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11216 (arc_loop_hazard): Remove function, functionality moved into
11218 (arc_hazard): Remove arc_loop_hazard call.
11219 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11220 into hwloop_optimize.
11221 (arc_label_align): Remove ZOL handling.
11222 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11223 * config/arc/arc.md (doloop_begin): Remove pattern.
11224 (doloop_begin_i): Likewise.
11225 (doloop_end_i): Likewise.
11226 (doloop_fallback): Likewise.
11227 (doloop_fallback_m): Likewise.
11228 (doloop_end): Reimplement expand.
11229 (arc_lp): New pattern for LP instruction.
11230 (loop_end): New pattern.
11231 (loop_fail): Likewise.
11232 (decrement_and_branch_until_zero): Likewise.
11233 * config/arc/arc.opt (mlpc-width): New option.
11234 * doc/invoke.texi (mlpc-width): Document option.
11236 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11238 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11239 (arc_ccfsm_advance): Fix checking for delay slots.
11240 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11242 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11244 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11246 (movhi_insn): Update store instruction constraint which are saving
11247 6-bit short immediates.
11248 (movsi_insn): Consider also short scaled load operations.
11249 (zero_extendhisi2_i): Use Usd constraint instead of T.
11250 (extendhisi2_i): Add q constraint.
11251 (arc_clzsi2): Add type and length attributes.
11252 (arc_ctzsi2): Likewise.
11253 * config/arc/constraints.md (Usc): Update constraint, the
11254 assembler can parse two relocations for a single instruction.
11256 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11258 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11259 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11261 2017-08-31 Olivier Hainque <hainque@adacore.com>
11263 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11264 match as powerpc-wrs-vxworks*.
11266 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11268 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11269 register constraint for by-element operand.
11270 (aarch64_mls_elt_merge<mode>): Likewise.
11272 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11274 * config/arc/arc.c (arc_can_follow_jump): Check for short
11277 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11279 * config.gcc: Use g.opt for arc.
11280 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11281 functionality moved to ...
11282 (legitimate_scaled_address_p): New function, ...here.
11283 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11284 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11285 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11287 (arc_override_options): Handle G option.
11288 (arc_output_pic_addr_const): Correct function definition.
11289 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11290 (arc_decl_anon_ns_mem_p): Delete.
11291 (arc_in_small_data_p): Overhaul this function to take into
11292 consideration the value given via G option.
11293 (arc_rewrite_small_data_1): Renamed and corrected old
11294 arc_rewrite_small_data function.
11295 (arc_rewrite_small_data): New function.
11296 (small_data_pattern): Don't use pic_offset_table_rtx.
11297 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11298 * config/arc/simdext.md (movmisalignv2hi): Use
11299 prepare_move_operands function.
11301 (movmisalign*): Likewise.
11302 * doc/invoke.texi (ARC options): Document -G option.
11304 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11306 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11308 * config/arc/arc.c (arc_print_operand): Output scalled address for
11309 sdata whenever is possible.
11310 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11311 load/stores are available.
11312 (compact_sda_memory_operand): Check for the alignment required by
11313 code density instructions.
11314 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11316 * config/arc/constraints.md (Usd): Update constraint.
11317 (Us0): New constraint.
11318 (Usc): Update constraint.
11320 2017-08-31 Richard Biener <rguenther@suse.de>
11322 PR middle-end/82054
11323 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11324 function only once.
11326 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11328 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11329 Resize type_signature.
11331 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11332 Alan Hayward <alan.hayward@arm.com>
11333 David Sherwood <david.sherwood@arm.com>
11335 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11336 subregs whose inner modes can be stored in GPRs.
11337 (aarch64_classify_index): Likewise.
11339 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11340 Alan Hayward <alan.hayward@arm.com>
11341 David Sherwood <david.sherwood@arm.com>
11343 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11344 (V_INT_EQUIV): ...this.
11345 (v_cmp_result): Rename to...
11346 (v_int_equiv): ...this.
11347 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11348 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11349 (copysign<mode>3): Likewise.
11350 (aarch64_simd_bsl<mode>_internal): Likewise.
11351 (aarch64_simd_bsl<mode>): Likewise.
11352 (vec_cmp<mode><mode>): Likewise.
11353 (vcond<mode><mode>): Likewise.
11354 (vcond<v_cmp_mixed><mode>): Likewise.
11355 (vcondu<mode><v_cmp_mixed>): Likewise.
11356 (aarch64_cm<optab><mode>): Likewise.
11357 (aarch64_cmtst<mode>): Likewise.
11358 (aarch64_fac<optab><mode>): Likewise.
11359 (vec_perm_const<mode>): Likewise.
11360 (vcond_mask_<mode><v_cmp_result>): Rename to...
11361 (vcond_mask_<mode><v_int_equiv>): ...this.
11362 (vec_cmp<mode><v_cmp_result>): Rename to...
11363 (vec_cmp<mode><v_int_equiv>): ...this.
11365 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11371 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11372 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11373 (UNSPEC_LD4_DREG): New unspecs.
11374 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11375 (aarch64_ld2<mode>_dreg_be): Replace with...
11376 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11378 (aarch64_ld3<mode>_dreg_le)
11379 (aarch64_ld3<mode>_dreg_be): Replace with...
11380 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11382 (aarch64_ld4<mode>_dreg_le)
11383 (aarch64_ld4<mode>_dreg_be): Replace with...
11384 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11387 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11389 PR tree-optimization/81987
11390 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11391 insert an initializer in a location not dominated by the stride
11394 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11396 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11397 on the entire header of the finally block in the fallthru case.
11399 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11401 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11403 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11405 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11406 rs6000_emit_move_from_cr and call renamed function.
11407 (rs6000_emit_prologue): Call renamed functions.
11408 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11409 movesi_from_cr, remove volatile CRs.
11411 2017-08-30 Jon Beniston <jon@beniston.com>
11412 Richard Biener <rguenther@suse.de>
11414 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11415 of VECTOR_MODE_P check.
11416 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11417 element vector types.
11419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11421 * df.h (df_read_modify_subreg_p): Remove in favor of...
11422 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11423 const_rtx instead of an rtx.
11424 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11425 * df-problems.c (df_word_lr_mark_ref): Likewise.
11426 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11427 (mark_pseudo_reg_dead): Likewise.
11428 (mark_ref_dead): Likewise.
11429 * reginfo.c (init_subregs_of_mode): Likewise.
11430 * sched-deps.c (sched_analyze_1): Likewise.
11431 * df-scan.c (df_def_record_1): Likewise.
11432 (df_uses_record): Likewise.
11433 (df_read_modify_subreg_p): Remove in favor of...
11434 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11435 const_rtx instead of an rtx.
11437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11438 Alan Hayward <alan.hayward@arm.com>
11439 David Sherwood <david.sherwood@arm.com>
11441 * rtl.h (partial_subreg_p): New function.
11442 * caller-save.c (save_call_clobbered_regs): Use it.
11443 * calls.c (expand_call): Likewise.
11444 * combine.c (combinable_i3pat): Likewise.
11445 (simplify_set): Likewise.
11446 (make_extraction): Likewise.
11447 (make_compound_operation_int): Likewise.
11448 (gen_lowpart_or_truncate): Likewise.
11449 (force_to_mode): Likewise.
11450 (make_field_assignment): Likewise.
11451 (reg_truncated_to_mode): Likewise.
11452 (record_truncated_value): Likewise.
11453 (move_deaths): Likewise.
11454 * cse.c (record_jump_cond): Likewise.
11455 (cse_insn): Likewise.
11456 * cselib.c (cselib_lookup_1): Likewise.
11457 * expmed.c (extract_bit_field_using_extv): Likewise.
11458 * function.c (assign_parm_setup_reg): Likewise.
11459 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11460 * ira-build.c (create_insn_allocnos): Likewise.
11461 * lra-coalesce.c (merge_pseudos): Likewise.
11462 * lra-constraints.c (match_reload): Likewise.
11463 (simplify_operand_subreg): Likewise.
11464 (curr_insn_transform): Likewise.
11465 * lra-lives.c (process_bb_lives): Likewise.
11466 * lra.c (new_insn_reg): Likewise.
11467 (lra_substitute_pseudo): Likewise.
11468 * regcprop.c (mode_change_ok): Likewise.
11469 (maybe_mode_change): Likewise.
11470 (copyprop_hardreg_forward_1): Likewise.
11471 * reload.c (push_reload): Likewise.
11472 (find_reloads): Likewise.
11473 (find_reloads_subreg_address): Likewise.
11474 * reload1.c (alter_reg): Likewise.
11475 (eliminate_regs_1): Likewise.
11476 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11478 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11480 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11481 back to if statements, including unpack.
11483 2017-08-30 Martin Liska <mliska@suse.cz>
11485 PR inline-asm/82001
11486 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11488 (func_checker::compare_asm_inputs_outputs): ... this function.
11489 (func_checker::compare_gimple_asm): Use the function to compare
11490 also ASM constrains.
11491 * ipa-icf-gimple.h: Rename the function.
11493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11494 Alan Hayward <alan.hayward@arm.com>
11495 David Sherwood <david.sherwood@arm.com>
11497 * coretypes.h (complex_mode): New type.
11498 * gdbhooks.py (build_pretty_printer): Handle it.
11499 * machmode.h (complex_mode): New class.
11500 (complex_mode::includes_p): New function.
11501 (is_complex_int_mode): Likewise.
11502 (is_complex_float_mode): Likewise.
11503 * genmodes.c (get_mode_class): Handle complex mode classes.
11504 * function.c (expand_function_end): Use is_complex_int_mode.
11506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11507 Alan Hayward <alan.hayward@arm.com>
11508 David Sherwood <david.sherwood@arm.com>
11510 * coretypes.h (scalar_mode_pod): New typedef.
11511 * gdbhooks.py (build_pretty_printer): Handle it.
11512 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11513 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11514 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11515 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11516 as_a <scalar_mode>.
11518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11519 Alan Hayward <alan.hayward@arm.com>
11520 David Sherwood <david.sherwood@arm.com>
11522 * machmode.h (mode_for_vector): Take a scalar_mode instead
11524 * stor-layout.c (mode_for_vector): Likewise.
11525 * explow.c (promote_mode): Use as_a <scalar_mode>.
11526 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11529 Alan Hayward <alan.hayward@arm.com>
11530 David Sherwood <david.sherwood@arm.com>
11532 * target.def (preferred_simd_mode): Take a scalar_mode
11533 instead of a machine_mode.
11534 * targhooks.h (default_preferred_simd_mode): Likewise.
11535 * targhooks.c (default_preferred_simd_mode): Likewise.
11536 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11537 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11538 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11539 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11540 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11541 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11542 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11543 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11545 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11546 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11547 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11548 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11549 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11550 * doc/tm.texi: Regenerate.
11551 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11555 Alan Hayward <alan.hayward@arm.com>
11556 David Sherwood <david.sherwood@arm.com>
11558 * target.def (scalar_mode_supported_p): Take a scalar_mode
11559 instead of a machine_mode.
11560 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11561 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11562 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11563 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11564 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11565 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11566 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11567 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11568 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11569 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11570 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11571 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11572 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11573 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11575 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11576 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11577 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11578 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11579 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11581 * doc/tm.texi: Regenerate.
11583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11584 Alan Hayward <alan.hayward@arm.com>
11585 David Sherwood <david.sherwood@arm.com>
11587 * coretypes.h (opt_scalar_mode): New typedef.
11588 * gdbhooks.py (build_pretty_printers): Handle it.
11589 * machmode.h (mode_iterator::get_2xwider): Add overload for
11591 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11593 * expr.c (convert_mode_scalar): Likewise.
11594 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11595 * optabs.c (expand_float): Likewise.
11596 (expand_fix): Likewise.
11597 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11600 Alan Hayward <alan.hayward@arm.com>
11601 David Sherwood <david.sherwood@arm.com>
11603 * optabs.c (expand_float): Explicitly check for scalars before
11604 using a branching expansion.
11605 (expand_fix): Likewise.
11607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11608 Alan Hayward <alan.hayward@arm.com>
11609 David Sherwood <david.sherwood@arm.com>
11611 * expr.c (convert_mode): Split scalar handling out into...
11612 (convert_mode_scalar): ...this new function. Treat the modes
11615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11616 Alan Hayward <alan.hayward@arm.com>
11617 David Sherwood <david.sherwood@arm.com>
11619 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11621 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11624 Alan Hayward <alan.hayward@arm.com>
11625 David Sherwood <david.sherwood@arm.com>
11627 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11628 rather than a machine_mode.
11629 (fixed_from_string): Likewise.
11630 (fixed_convert): Likewise.
11631 (fixed_convert_from_int): Likewise.
11632 (fixed_convert_from_real): Likewise.
11633 (real_convert_from_fixed): Likewise.
11634 * fixed-value.c (fixed_from_double_int): Likewise.
11635 (fixed_from_string): Likewise.
11636 (fixed_convert): Likewise.
11637 (fixed_convert_from_int): Likewise.
11638 (fixed_convert_from_real): Likewise.
11639 (real_convert_from_fixed): Likewise.
11640 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11643 Alan Hayward <alan.hayward@arm.com>
11644 David Sherwood <david.sherwood@arm.com>
11646 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11647 of separate mode class checks. Do not allow vector modes here.
11648 (immed_wide_int_const): Use as_a <scalar_mode>.
11649 * explow.c (trunc_int_for_mode): Likewise.
11650 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11651 (wi::shwi): Likewise.
11652 (wi::min_value): Likewise.
11653 (wi::max_value): Likewise.
11654 * dwarf2out.c (loc_descriptor): Likewise.
11655 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11656 for CONST_WIDE_INT.
11658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11659 Alan Hayward <alan.hayward@arm.com>
11660 David Sherwood <david.sherwood@arm.com>
11662 * tree.h (SCALAR_TYPE_MODE): New macro.
11663 * expr.c (expand_expr_addr_expr_1): Use it.
11664 (expand_expr_real_2): Likewise.
11665 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11666 (fold_convert_const_fixed_from_int): Likewise.
11667 (fold_convert_const_fixed_from_real): Likewise.
11668 (native_encode_fixed): Likewise
11669 (native_encode_complex): Likewise
11670 (native_encode_vector): Likewise.
11671 (native_interpret_fixed): Likewise.
11672 (native_interpret_real): Likewise.
11673 (native_interpret_complex): Likewise.
11674 (native_interpret_vector): Likewise.
11675 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11676 (simd_clone_adjust_argument_types): Likewise.
11677 (simd_clone_init_simd_arrays): Likewise.
11678 (simd_clone_adjust): Likewise.
11679 * stor-layout.c (layout_type): Likewise.
11680 * tree.c (build_minus_one_cst): Likewise.
11681 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11682 * tree-inline.c (estimate_move_cost): Likewise.
11683 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11684 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11685 (vectorizable_reduction): Likewise.
11686 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11687 (vect_recog_mixed_size_cond_pattern): Likewise.
11688 (check_bool_pattern): Likewise.
11689 (adjust_bool_pattern): Likewise.
11690 (search_type_for_mask_1): Likewise.
11691 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11692 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11693 (vectorizable_load): Likewise.
11694 (vectorizable_store): Likewise.
11695 * ubsan.c (ubsan_encode_value): Likewise.
11696 * varasm.c (output_constant): Likewise.
11698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11699 Alan Hayward <alan.hayward@arm.com>
11700 David Sherwood <david.sherwood@arm.com>
11702 * coretypes.h (scalar_mode): New class.
11703 * machmode.h (scalar_mode): Likewise.
11704 (scalar_mode::includes_p): New function.
11705 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11706 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11707 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11708 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11709 * expmed.c (store_bit_field_1): Likewise.
11710 (extract_bit_field_1): Likewise.
11711 * expr.c (write_complex_part): Likewise.
11712 (read_complex_part): Likewise.
11713 (emit_move_complex_push): Likewise.
11714 (expand_expr_real_2): Likewise.
11715 * function.c (assign_parm_setup_reg): Likewise.
11716 (assign_parms_unsplit_complex): Likewise.
11717 * optabs.c (expand_binop): Likewise.
11718 * rtlanal.c (subreg_get_info): Likewise.
11719 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11720 * varasm.c (output_constant_pool_2): Likewise.
11722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11723 Alan Hayward <alan.hayward@arm.com>
11724 David Sherwood <david.sherwood@arm.com>
11726 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11728 (expmed_mult_highpart_optab): Likewise.
11729 (expmed_mult_highpart): Likewise.
11731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11732 Alan Hayward <alan.hayward@arm.com>
11733 David Sherwood <david.sherwood@arm.com>
11735 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11736 instead of a machine_mode.
11737 (builtin_memset_read_str): Likewise.
11738 * builtins.c (c_readstr): Likewise.
11739 (builtin_memcpy_read_str): Likewise.
11740 (builtin_strncpy_read_str): Likewise.
11741 (builtin_memset_read_str): Likewise.
11742 (builtin_memset_gen_str): Likewise.
11743 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11744 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11745 instead of a machine_mode.
11746 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11748 (make_extraction): Likewise.
11749 (try_widen_shift_mode): Take and return scalar_int_modes instead
11751 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11752 a scalar_int_mode instead of a machine_mode.
11753 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11754 (avr_addr_space_pointer_mode): Likewise.
11755 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11756 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11757 (msp430_unwind_word_mode): Likewise.
11758 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11759 (spu_addr_space_pointer_mode): Likewise.
11760 (spu_addr_space_address_mode): Likewise.
11761 (spu_libgcc_cmp_return_mode): Likewise.
11762 (spu_libgcc_shift_count_mode): Likewise.
11763 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11764 (rl78_addr_space_pointer_mode): Likewise.
11765 (fl78_unwind_word_mode): Likewise.
11766 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11768 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11769 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11770 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11771 (mips_valid_pointer_mode): Likewise.
11772 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11773 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11774 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11776 (ft32_addr_space_address_mode): Likewise.
11777 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11778 scalar_int_mode instead of a machine_mode.
11779 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11781 (m32c_addr_space_address_mode): Likewise.
11782 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11783 (rs6000_eh_return_filter_mode): Likewise.
11784 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11785 (rs6000_eh_return_filter_mode): Likewise.
11786 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11787 (s390_libgcc_shift_count_mode): Likewise.
11788 (s390_unwind_word_mode): Likewise.
11789 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11791 * target.def (mode_rep_extended): Likewise.
11792 (valid_pointer_mode): Likewise.
11793 (addr_space.valid_pointer_mode): Likewise.
11794 (eh_return_filter_mode): Return a scalar_int_mode rather than
11796 (libgcc_cmp_return_mode): Likewise.
11797 (libgcc_shift_count_mode): Likewise.
11798 (unwind_word_mode): Likewise.
11799 (addr_space.pointer_mode): Likewise.
11800 (addr_space.address_mode): Likewise.
11801 * doc/tm.texi: Regenerate.
11802 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11804 (do_jump): Use scalar_int_mode for local variables.
11805 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11806 rather than a machine_mode.
11807 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11808 (scompare_loc_descriptor_wide): Likewise.
11809 (scompare_loc_descriptor_narrow): Likewise.
11810 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11812 * except.c (sjlj_emit_dispatch_table): Likewise.
11813 (expand_builtin_eh_copy_values): Likewise.
11814 * explow.c (convert_memory_address_addr_space_1): Likewise.
11815 Take a scalar_int_mode rather than a machine_mode.
11816 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11817 than a machine_mode.
11818 (memory_address_addr_space): Use scalar_int_mode for local variables.
11819 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11820 rather than a machine_mode.
11821 * expmed.c (mask_rtx): Likewise.
11822 (init_expmed_one_conv): Likewise.
11823 (expand_mult_highpart_adjust): Likewise.
11824 (extract_high_half): Likewise.
11825 (expmed_mult_highpart_optab): Likewise.
11826 (expmed_mult_highpart): Likewise.
11827 (expand_smod_pow2): Likewise.
11828 (expand_sdiv_pow2): Likewise.
11829 (emit_store_flag_int): Likewise.
11830 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11832 (extract_low_bits): Likewise.
11833 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11835 * expr.c (pieces_addr::adjust): Likewise.
11836 (can_store_by_pieces): Likewise.
11837 (store_by_pieces): Likewise.
11838 (clear_by_pieces_1): Likewise.
11839 (expand_expr_addr_expr_1): Likewise.
11840 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11841 (expand_expr_real_1): Likewise.
11842 (try_casesi): Likewise.
11843 * final.c (shorten_branches): Likewise.
11844 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11845 type of "mode" to machine_mode.
11846 * internal-fn.c (expand_arith_overflow_result_store): Take a
11847 scalar_int_mode rather than a machine_mode.
11848 (expand_mul_overflow): Use scalar_int_mode for local variables.
11849 * loop-doloop.c (doloop_modify): Likewise.
11850 (doloop_optimize): Likewise.
11851 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11852 than a machine_mode.
11853 (expand_doubleword_shift_condmove): Likewise.
11854 (expand_doubleword_shift): Likewise.
11855 (expand_doubleword_clz): Likewise.
11856 (expand_doubleword_popcount): Likewise.
11857 (expand_doubleword_parity): Likewise.
11858 (expand_absneg_bit): Use scalar_int_mode for local variables.
11859 (prepare_float_lib_cmp): Likewise.
11860 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11861 rather than a machine_mode.
11862 (convert_memory_address_addr_space): Likewise.
11863 (get_mode_bounds): Likewise.
11864 (get_address_mode): Return a scalar_int_mode rather than a
11866 * rtlanal.c (get_address_mode): Likewise.
11867 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11868 than a machine_mode.
11869 * targhooks.c (default_mode_rep_extended): Likewise.
11870 (default_valid_pointer_mode): Likewise.
11871 (default_addr_space_valid_pointer_mode): Likewise.
11872 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11873 than a machine_mode.
11874 (default_libgcc_cmp_return_mode): Likewise.
11875 (default_libgcc_shift_count_mode): Likewise.
11876 (default_unwind_word_mode): Likewise.
11877 (default_addr_space_pointer_mode): Likewise.
11878 (default_addr_space_address_mode): Likewise.
11879 * targhooks.h (default_eh_return_filter_mode): Likewise.
11880 (default_libgcc_cmp_return_mode): Likewise.
11881 (default_libgcc_shift_count_mode): Likewise.
11882 (default_unwind_word_mode): Likewise.
11883 (default_addr_space_pointer_mode): Likewise.
11884 (default_addr_space_address_mode): Likewise.
11885 (default_mode_rep_extended): Take a scalar_int_mode rather than
11887 (default_valid_pointer_mode): Likewise.
11888 (default_addr_space_valid_pointer_mode): Likewise.
11889 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11891 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11892 rather than a machine_mode.
11893 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11894 for local variables.
11895 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11896 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11897 than a machine_mode.
11899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11900 Alan Hayward <alan.hayward@arm.com>
11901 David Sherwood <david.sherwood@arm.com>
11903 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11904 the mode argument to scalar_int_mode.
11905 (do_jump_by_parts_zero_rtx): Likewise.
11906 (do_jump_by_parts_equality_rtx): Likewise.
11907 (do_jump_by_parts_greater): Take a mode argument.
11908 (do_jump_by_parts_equality): Likewise.
11909 (do_jump_1): Update calls accordingly.
11911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11912 Alan Hayward <alan.hayward@arm.com>
11913 David Sherwood <david.sherwood@arm.com>
11915 * is-a.h (safe_dyn_cast): New function.
11916 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11917 (jump_table_for_label): Likewise.
11918 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11919 instead of an rtx_insn *.
11920 (shorten_branches): Use dyn_cast instead of LABEL_P and
11921 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11922 rtx_jump_table_data::get_data_mode.
11923 (final_scan_insn): Likewise.
11925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11926 Alan Hayward <alan.hayward@arm.com>
11927 David Sherwood <david.sherwood@arm.com>
11929 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11930 trying to combine a full-register integer set with a subreg
11933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11934 Alan Hayward <alan.hayward@arm.com>
11935 David Sherwood <david.sherwood@arm.com>
11937 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11938 that is always either address_mode or pointer_mode.
11940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11941 Alan Hayward <alan.hayward@arm.com>
11942 David Sherwood <david.sherwood@arm.com>
11944 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11945 when the two are known to be equal.
11947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11948 Alan Hayward <alan.hayward@arm.com>
11949 David Sherwood <david.sherwood@arm.com>
11951 * simplify-rtx.c (simplify_const_unary_operation): Use
11952 is_a <scalar_int_mode> instead of checking for a nonzero
11953 precision. Forcibly convert op_mode to a scalar_int_mode
11954 in that case. More clearly differentiate the operand and
11955 result modes and use the former when deciding what the value
11956 of a count-bits operation should be. Use is_int_mode instead
11957 of checking for a MODE_INT. Remove redundant check for whether
11958 this mode has a zero precision.
11960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11961 Alan Hayward <alan.hayward@arm.com>
11962 David Sherwood <david.sherwood@arm.com>
11964 * optabs.c (widen_leading): Change the type of the mode argument
11965 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11966 (widen_bswap): Likewise.
11967 (expand_parity): Likewise.
11968 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11969 (expand_ffs): Likewise.
11970 (epand_unop): Check for scalar integer modes before calling the
11973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11974 Alan Hayward <alan.hayward@arm.com>
11975 David Sherwood <david.sherwood@arm.com>
11977 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11979 (expand_expr_real_1): Update call accordingly.
11981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11982 Alan Hayward <alan.hayward@arm.com>
11983 David Sherwood <david.sherwood@arm.com>
11985 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11986 value_mode arguments. Use scalar_int_mode internally.
11987 (store_bit_field_1): Rename the new integer mode from imode
11988 to op0_mode and use it instead of GET_MODE (op0). Update calls
11989 to store_split_bit_field, store_bit_field_using_insv and
11990 store_fixed_bit_field.
11991 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11992 Use scalar_int_mode internally. Use a bit count rather than a mode
11993 when calculating the largest bit size for get_best_mode.
11994 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11995 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11996 Remove assertion that OP0 has a scalar integer mode.
11997 (store_split_bit_field): Add op0_mode and value_mode arguments.
11998 Update calls to extract_fixed_bit_field.
11999 (extract_bit_field_using_extv): Add an op0_mode argument.
12000 Use scalar_int_mode internally.
12001 (extract_bit_field_1): Rename the new integer mode from imode to
12002 op0_mode and use it instead of GET_MODE (op0). Update calls to
12003 extract_split_bit_field, extract_bit_field_using_extv and
12004 extract_fixed_bit_field.
12005 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12006 to extract_split_bit_field and extract_fixed_bit_field_1.
12007 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12008 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12009 on the target mode.
12010 (extract_split_bit_field): Add an op0_mode argument. Update call
12011 to extract_fixed_bit_field.
12013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12014 Alan Hayward <alan.hayward@arm.com>
12015 David Sherwood <david.sherwood@arm.com>
12017 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12018 * explow.c (hard_function_value): Likewise.
12019 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12020 convert_to_mode call outside the loop.
12021 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12022 for the mode iterator. Require the mode specified by max_pieces
12024 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12026 (copy_blkmode_to_reg): Likewise.
12027 (set_storage_via_setmem): Likewise.
12028 * optabs.c (prepare_cmp_insn): Likewise.
12029 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12030 * stor-layout.c (finish_bitfield_representative): Likewise.
12032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12033 Alan Hayward <alan.hayward@arm.com>
12034 David Sherwood <david.sherwood@arm.com>
12036 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12037 * expr.c (convert_move): Use them.
12038 (convert_modes): Likewise.
12039 (store_expr_with_bounds): Likewise.
12041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12042 Alan Hayward <alan.hayward@arm.com>
12043 David Sherwood <david.sherwood@arm.com>
12045 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12046 parameter for the mode of "x". Remove the "known_x", "known_mode"
12047 and "known_ret" arguments. Change the type of the mode argument
12048 to scalar_int_mode.
12049 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12050 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12051 (reg_num_sign_bit_copies_for_combine): Likewise.
12052 * rtlanal.c (nonzero_bits1): Likewise.
12053 (num_sign_bit_copies1): Likewise.
12054 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12055 (reg_num_sign_bit_copies_general): Likewise.
12056 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12057 (reg_nonzero_bits_general): Likewise.
12059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12060 Alan Hayward <alan.hayward@arm.com>
12061 David Sherwood <david.sherwood@arm.com>
12063 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12064 than in subroutines. Return 1 for non-integer modes.
12065 (cached_num_sign_bit_copies): Change the type of the mode parameter
12066 to scalar_int_mode.
12067 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12068 classes. Handle CONST_INT_P first and then check whether X also
12069 has a scalar integer mode. Check the same thing for inner registers
12070 of a SUBREG and for values that are being extended or truncated.
12072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12073 Alan Hayward <alan.hayward@arm.com>
12074 David Sherwood <david.sherwood@arm.com>
12076 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12077 in subroutines. Return the mode mask for non-integer modes.
12078 (cached_nonzero_bits): Change the type of the mode parameter
12079 to scalar_int_mode.
12080 (nonzero_bits1): Likewise. Remove early exit for other mode
12081 classes. Handle CONST_INT_P first and then check whether X
12082 also has a scalar integer mode.
12084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12085 Alan Hayward <alan.hayward@arm.com>
12086 David Sherwood <david.sherwood@arm.com>
12088 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12089 Return a scalar_int_mode and assert that the size is greater than
12091 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12092 (op_by_pieces_d::op_by_pieces_d): Likewise.
12093 (op_by_pieces_d::run): Likewise.
12094 (can_store_by_pieces): Likewise.
12096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12097 Alan Hayward <alan.hayward@arm.com>
12098 David Sherwood <david.sherwood@arm.com>
12100 * combine.c (extract_left_shift): Add a mode argument and update
12102 (make_compound_operation_int): Change the type of the mode parameter
12103 to scalar_int_mode and update the call to extract_left_shift.
12105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12106 Alan Hayward <alan.hayward@arm.com>
12107 David Sherwood <david.sherwood@arm.com>
12109 * combine.c (simplify_and_const_int): Change the type of the mode
12110 parameter to scalar_int_mode.
12111 (simplify_and_const_int_1): Likewise. Update recursive call.
12113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12114 Alan Hayward <alan.hayward@arm.com>
12115 David Sherwood <david.sherwood@arm.com>
12117 * combine.c (simplify_compare_const): Check that the mode is a
12118 scalar_int_mode (rather than VOIDmode) before testing its
12120 (simplify_comparison): Move COMPARISON_P handling out of the
12121 loop and restrict the latter part of the loop to scalar_int_modes.
12122 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12123 and when considering SUBREG_REGs. Use is_int_mode instead of
12124 checking GET_MODE_CLASS against MODE_INT.
12126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12127 Alan Hayward <alan.hayward@arm.com>
12128 David Sherwood <david.sherwood@arm.com>
12130 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12131 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12132 shift_unit_mode and for modes involved in scalar shifts.
12134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12135 Alan Hayward <alan.hayward@arm.com>
12136 David Sherwood <david.sherwood@arm.com>
12138 * combine.c (force_int_to_mode): New function, split out from...
12139 (force_to_mode): ...here. Keep xmode up-to-date and use it
12140 instead of GET_MODE (x).
12142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12143 Alan Hayward <alan.hayward@arm.com>
12144 David Sherwood <david.sherwood@arm.com>
12146 * optabs-query.h (extraction_insn::struct_mode): Change type to
12147 opt_scalar_int_mode and update comment.
12148 (extraction_insn::field_mode): Change type to scalar_int_mode.
12149 (extraction_insn::pos_mode): Likewise.
12150 * combine.c (make_extraction): Update accordingly.
12151 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12152 (get_optab_extraction_insn): Likewise.
12153 * recog.c (simplify_while_replacing): Likewise.
12154 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12155 parameter to opt_scalar_int_mode.
12157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12158 Alan Hayward <alan.hayward@arm.com>
12159 David Sherwood <david.sherwood@arm.com>
12161 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12162 to a scalar_int_mode instead of a machine_mode.
12163 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12164 (get_best_mode): Return a boolean and use a pointer argument to store
12165 the selected mode. Replace the limit mode parameter with a bit limit.
12166 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12167 for the values returned by bit_field_mode_iterator::next_mode.
12168 (store_bit_field): Update call to get_best_mode.
12169 (store_fixed_bit_field): Likewise.
12170 (extract_fixed_bit_field): Likewise.
12171 * expr.c (optimize_bitfield_assignment_op): Likewise.
12172 * fold-const.c (optimize_bit_field_compare): Likewise.
12173 (fold_truth_andor_1): Likewise.
12174 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12175 Update for new type of m_mode.
12176 (get_best_mode): As above.
12178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12179 Alan Hayward <alan.hayward@arm.com>
12180 David Sherwood <david.sherwood@arm.com>
12182 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12183 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12184 (store_bit_field): Check is_a <scalar_int_mode> before calling
12185 strict_volatile_bitfield_p.
12186 (extract_bit_field): Likewise.
12188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12189 Alan Hayward <alan.hayward@arm.com>
12190 David Sherwood <david.sherwood@arm.com>
12192 * target.def (cstore_mode): Return a scalar_int_mode.
12193 * doc/tm.texi: Regenerate.
12194 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12195 * targhooks.h (default_cstore_mode): Likewise.
12196 * targhooks.c (default_cstore_mode): Likewise, using a forced
12198 * expmed.c (emit_cstore): Expect the target of the cstore to be
12201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12202 Alan Hayward <alan.hayward@arm.com>
12203 David Sherwood <david.sherwood@arm.com>
12205 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12207 (niter_desc): Likewise mode.
12208 (iv_analyze): Add a mode parameter.
12210 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12211 and change its type to scalar_int_mode.
12212 * loop-iv.c: Update commentary at head of file.
12213 (iv_constant): Pass the mode paraeter before the rtx it describes
12214 and change its type to scalar_int_mode. Remove VOIDmode handling.
12215 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12216 (iv_extend): Likewise.
12217 (shorten_into_mode): Likewise.
12218 (iv_add): Use scalar_int_mode.
12219 (iv_mult): Likewise.
12220 (iv_shift): Likewise.
12221 (canonicalize_iv_subregs): Likewise.
12222 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12223 it describes and change its mode to scalar_int_mode. Also change
12224 the type of the returned inner_mode to scalar_int_mode.
12225 (get_biv_step): Likewise, turning outer_mode from a pointer
12226 into a direct parameter. Update call to get_biv_step_1.
12227 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12228 iv_constant and get_biv_step.
12229 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12230 and change its type to scalar_int_mode. Don't initialise iv->mode
12231 to VOIDmode and remove later checks for its still being VOIDmode.
12232 Update calls to iv_analyze_op and iv_analyze_expr. Check
12233 is_a <scalar_int_mode> when changing the mode under consideration.
12234 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12235 Update call to iv_analyze_expr.
12236 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12237 inner register is not also a scalar_int_mode. Update call to
12239 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12240 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12241 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12242 separate mode class checks. Update calls to iv_analyze. Remove
12243 fix-up of VOIDmodes after iv_analyze_biv.
12244 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12245 don't have a scalar_int_mode. Update call to biv_p.
12247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12248 Alan Hayward <alan.hayward@arm.com>
12249 David Sherwood <david.sherwood@arm.com>
12251 * cfgexpand.c (convert_debug_memory_address): Use
12252 as_a <scalar_int_mode>.
12253 * combine.c (expand_compound_operation): Likewise.
12254 (make_extraction): Likewise.
12255 (change_zero_ext): Likewise.
12256 (simplify_comparison): Likewise.
12257 * cse.c (cse_insn): Likewise.
12258 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12259 (mem_loc_descriptor): Likewise.
12260 (loc_descriptor): Likewise.
12261 * expmed.c (init_expmed_one_mode): Likewise.
12262 (synth_mult): Likewise.
12263 (emit_store_flag_1): Likewise.
12264 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12265 of a comparison with size.
12266 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12267 (reduce_to_bit_field_precision): Likewise.
12268 * function.c (expand_function_end): Likewise.
12269 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12270 * loop-doloop.c (doloop_modify): Likewise.
12271 * optabs.c (expand_binop): Likewise.
12272 (expand_unop): Likewise.
12273 (expand_copysign_absneg): Likewise.
12274 (prepare_cmp_insn): Likewise.
12275 (maybe_legitimize_operand): Likewise.
12276 * recog.c (const_scalar_int_operand): Likewise.
12277 * rtlanal.c (get_address_mode): Likewise.
12278 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12279 (simplify_cond_clz_ctz): Likewise.
12280 * tree-nested.c (get_nl_goto_field): Likewise.
12281 * tree.c (build_vector_type_for_mode): Likewise.
12282 * var-tracking.c (use_narrower_mode): Likewise.
12284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12285 Alan Hayward <alan.hayward@arm.com>
12286 David Sherwood <david.sherwood@arm.com>
12288 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12289 * builtins.c (expand_builtin_signbit): Use it.
12290 * cfgexpand.c (expand_debug_expr): Likewise.
12291 * dojump.c (do_jump): Likewise.
12292 (do_compare_and_jump): Likewise.
12293 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12294 * expmed.c (make_tree): Likewise.
12295 * expr.c (expand_expr_real_2): Likewise.
12296 (expand_expr_real_1): Likewise.
12297 (try_casesi): Likewise.
12298 * fold-const-call.c (fold_const_call_ss): Likewise.
12299 * fold-const.c (unextend): Likewise.
12300 (extract_muldiv_1): Likewise.
12301 (fold_single_bit_test): Likewise.
12302 (native_encode_int): Likewise.
12303 (native_encode_string): Likewise.
12304 (native_interpret_int): Likewise.
12305 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12306 * internal-fn.c (expand_addsub_overflow): Likewise.
12307 (expand_neg_overflow): Likewise.
12308 (expand_mul_overflow): Likewise.
12309 (expand_arith_overflow): Likewise.
12310 * match.pd: Likewise.
12311 * stor-layout.c (layout_type): Likewise.
12312 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12313 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12314 * tree-ssanames.c (get_range_info): Likewise.
12315 * tree-switch-conversion.c (array_value_type) Likewise.
12316 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12317 (vect_recog_divmod_pattern): Likewise.
12318 (vect_recog_mixed_size_cond_pattern): Likewise.
12319 * tree-vrp.c (extract_range_basic): Likewise.
12320 (simplify_float_conversion_using_ranges): Likewise.
12321 * tree.c (int_fits_type_p): Likewise.
12322 * ubsan.c (instrument_bool_enum_load): Likewise.
12323 * varasm.c (mergeable_string_section): Likewise.
12324 (narrowing_initializer_constant_valid_p): Likewise.
12325 (output_constant): Likewise.
12327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12328 Alan Hayward <alan.hayward@arm.com>
12329 David Sherwood <david.sherwood@arm.com>
12331 * machmode.h (NARROWEST_INT_MODE): New macro.
12332 * expr.c (alignment_for_piecewise_move): Use it instead of
12333 GET_CLASS_NARROWEST_MODE (MODE_INT).
12334 (push_block): Likewise.
12335 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12337 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12340 Alan Hayward <alan.hayward@arm.com>
12341 David Sherwood <david.sherwood@arm.com>
12343 * postreload.c (move2add_valid_value_p): Change the type of the
12344 mode parameter to scalar_int_mode.
12345 (move2add_use_add2_insn): Add a mode parameter and use it instead
12347 (move2add_use_add3_insn): Likewise.
12348 (reload_cse_move2add): Update accordingly.
12350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12351 Alan Hayward <alan.hayward@arm.com>
12352 David Sherwood <david.sherwood@arm.com>
12354 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12356 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12357 * optabs.c (expand_unop): Likewise.
12359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12360 Alan Hayward <alan.hayward@arm.com>
12361 David Sherwood <david.sherwood@arm.com>
12363 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12364 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12365 (popcount_loc_descriptor): Likewise.
12366 (bswap_loc_descriptor): Likewise.
12367 (rotate_loc_descriptor): Likewise.
12368 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12369 calling the functions above.
12371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12372 Alan Hayward <alan.hayward@arm.com>
12373 David Sherwood <david.sherwood@arm.com>
12375 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12377 (try_combine): Likewise.
12378 (simplify_if_then_else): Likewise.
12379 * cse.c (cse_insn): Likewise.
12380 * dwarf2out.c (mem_loc_descriptor): Likewise.
12381 * emit-rtl.c (gen_lowpart_common): Likewise.
12382 * simplify-rtx.c (simplify_truncation): Likewise.
12383 (simplify_binary_operation_1): Likewise.
12384 (simplify_const_relational_operation): Likewise.
12385 (simplify_ternary_operation): Likewise.
12386 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12389 Alan Hayward <alan.hayward@arm.com>
12390 David Sherwood <david.sherwood@arm.com>
12392 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12393 * reload.c (push_reload): Likewise.
12394 (find_reloads): Likewise.
12396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12397 Alan Hayward <alan.hayward@arm.com>
12398 David Sherwood <david.sherwood@arm.com>
12400 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12401 (make_compound_operation_int): Likewise.
12402 (change_zero_ext): Likewise.
12403 * expr.c (convert_move): Likewise.
12404 (convert_modes): Likewise.
12405 * fwprop.c (forward_propagate_subreg): Likewise.
12406 * loop-iv.c (get_biv_step_1): Likewise.
12407 * optabs.c (widen_operand): Likewise.
12408 * postreload.c (move2add_valid_value_p): Likewise.
12409 * recog.c (simplify_while_replacing): Likewise.
12410 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12411 (simplify_binary_operation_1): Likewise. Remove redundant
12412 mode equality check.
12414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12415 Alan Hayward <alan.hayward@arm.com>
12416 David Sherwood <david.sherwood@arm.com>
12418 * combine.c (combine_simplify_rtx): Add checks for
12419 is_a <scalar_int_mode>.
12420 (simplify_if_then_else): Likewise.
12421 (make_field_assignment): Likewise.
12422 (simplify_comparison): Likewise.
12423 * ifcvt.c (noce_try_bitop): Likewise.
12424 * loop-invariant.c (canonicalize_address_mult): Likewise.
12425 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12428 Alan Hayward <alan.hayward@arm.com>
12429 David Sherwood <david.sherwood@arm.com>
12431 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12432 is_a <scalar_int_mode> instead of != BLKmode.
12434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12435 Alan Hayward <alan.hayward@arm.com>
12436 David Sherwood <david.sherwood@arm.com>
12438 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12439 instead of != VOIDmode.
12440 * combine.c (if_then_else_cond): Likewise.
12441 (change_zero_ext): Likewise.
12442 * dwarf2out.c (mem_loc_descriptor): Likewise.
12443 (loc_descriptor): Likewise.
12444 * rtlanal.c (canonicalize_condition): Likewise.
12445 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12448 Alan Hayward <alan.hayward@arm.com>
12449 David Sherwood <david.sherwood@arm.com>
12451 * simplify-rtx.c (simplify_binary_operation_1): Use
12452 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12455 Alan Hayward <alan.hayward@arm.com>
12456 David Sherwood <david.sherwood@arm.com>
12458 * wide-int.h (int_traits<unsigned char>) New class.
12459 (int_traits<unsigned short>) Likewise.
12460 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12461 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12463 * combine.c (set_nonzero_bits_and_sign_copies): Use
12464 is_a <scalar_int_mode>.
12465 (find_split_point): Likewise.
12466 (combine_simplify_rtx): Likewise.
12467 (simplify_logical): Likewise.
12468 (expand_compound_operation): Likewise.
12469 (expand_field_assignment): Likewise.
12470 (make_compound_operation): Likewise.
12471 (extended_count): Likewise.
12472 (change_zero_ext): Likewise.
12473 (simplify_comparison): Likewise.
12474 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12475 (ucompare_loc_descriptor): Likewise.
12476 (minmax_loc_descriptor): Likewise.
12477 (mem_loc_descriptor): Likewise.
12478 (loc_descriptor): Likewise.
12479 * expmed.c (init_expmed_one_mode): Likewise.
12480 * lra-constraints.c (lra_constraint_offset): Likewise.
12481 * optabs.c (prepare_libcall_arg): Likewise.
12482 * postreload.c (move2add_note_store): Likewise.
12483 * reload.c (operands_match_p): Likewise.
12484 * rtl.h (load_extend_op): Likewise.
12485 * rtlhooks.c (gen_lowpart_general): Likewise.
12486 * simplify-rtx.c (simplify_truncation): Likewise.
12487 (simplify_unary_operation_1): Likewise.
12488 (simplify_binary_operation_1): Likewise.
12489 (simplify_const_binary_operation): Likewise.
12490 (simplify_const_relational_operation): Likewise.
12491 (simplify_subreg): Likewise.
12492 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12493 * var-tracking.c (adjust_mems): Likewise.
12494 (prepare_call_arguments): Likewise.
12496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12497 Alan Hayward <alan.hayward@arm.com>
12498 David Sherwood <david.sherwood@arm.com>
12500 * machmode.h (is_int_mode): New fuction.
12501 * combine.c (find_split_point): Use it.
12502 (combine_simplify_rtx): Likewise.
12503 (simplify_if_then_else): Likewise.
12504 (simplify_set): Likewise.
12505 (simplify_shift_const_1): Likewise.
12506 (simplify_comparison): Likewise.
12507 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12508 * cse.c (notreg_cost): Likewise.
12509 (cse_insn): Likewise.
12510 * cselib.c (cselib_lookup_1): Likewise.
12511 * dojump.c (do_jump_1): Likewise.
12512 (do_compare_rtx_and_jump): Likewise.
12513 * dse.c (get_call_args): Likewise.
12514 * dwarf2out.c (rtl_for_decl_init): Likewise.
12515 (native_encode_initializer): Likewise.
12516 * expmed.c (emit_store_flag_1): Likewise.
12517 (emit_store_flag): Likewise.
12518 * expr.c (convert_modes): Likewise.
12519 (store_field): Likewise.
12520 (expand_expr_real_1): Likewise.
12521 * fold-const.c (fold_read_from_constant_string): Likewise.
12522 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12523 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12524 * optabs.c (expand_binop): Likewise.
12525 (expand_unop): Likewise.
12526 (expand_abs_nojump): Likewise.
12527 (expand_one_cmpl_abs_nojump): Likewise.
12528 * simplify-rtx.c (mode_signbit_p): Likewise.
12529 (val_signbit_p): Likewise.
12530 (val_signbit_known_set_p): Likewise.
12531 (val_signbit_known_clear_p): Likewise.
12532 (simplify_relational_operation_1): Likewise.
12533 * tree.c (vector_type_mode): Likewise.
12535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12536 Alan Hayward <alan.hayward@arm.com>
12537 David Sherwood <david.sherwood@arm.com>
12539 * machmode.h (smallest_mode_for_size): Fix formatting.
12540 (smallest_int_mode_for_size): New function.
12541 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12542 instead of smallest_mode_for_size.
12543 * combine.c (make_extraction): Likewise.
12544 * config/arc/arc.c (arc_expand_movmem): Likewise.
12545 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12546 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12547 * config/s390/s390.c (s390_expand_insv): Likewise.
12548 * config/sparc/sparc.c (assign_int_registers): Likewise.
12549 * config/spu/spu.c (spu_function_value): Likewise.
12550 (spu_function_arg): Likewise.
12551 * coverage.c (get_gcov_type): Likewise.
12552 (get_gcov_unsigned_t): Likewise.
12553 * dse.c (find_shift_sequence): Likewise.
12554 * expmed.c (store_bit_field_1): Likewise.
12555 * expr.c (convert_move): Likewise.
12556 (store_field): Likewise.
12557 * internal-fn.c (expand_arith_overflow): Likewise.
12558 * optabs-query.c (get_best_extraction_insn): Likewise.
12559 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12560 * stor-layout.c (layout_type): Likewise.
12561 (initialize_sizetypes): Likewise.
12562 * targhooks.c (default_get_mask_mode): Likewise.
12563 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12566 Alan Hayward <alan.hayward@arm.com>
12567 David Sherwood <david.sherwood@arm.com>
12569 * machmode.h (opt_mode::else_blk): New function.
12570 (int_mode_for_mode): Declare.
12571 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12572 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12574 * cfgexpand.c (expand_debug_expr): Likewise.
12575 * combine.c (gen_lowpart_or_truncate): Likewise.
12576 (gen_lowpart_for_combine): Likewise.
12577 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12578 * config/avr/avr.c (avr_to_int_mode): Likewise.
12579 (avr_out_plus_1): Likewise.
12580 (avr_out_plus): Likewise.
12581 (avr_out_round): Likewise.
12582 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12583 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12584 (s390_expand_vcond): Likewise.
12585 * config/spu/spu.c (spu_split_immediate): Likewise.
12586 (spu_expand_mov): Likewise.
12587 * dse.c (get_stored_val): Likewise.
12588 * expmed.c (store_bit_field_1): Likewise.
12589 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12591 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12592 (extract_low_bits): Likewise.
12593 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12594 handling rather than repeating the check.
12595 (emit_group_store): Likewise.
12596 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12597 * optabs.c (expand_absneg_bit): Likewise.
12598 (expand_copysign_absneg): Likewise.
12599 (expand_copysign_bit): Likewise.
12600 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12601 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12602 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12603 * var-tracking.c (prepare_call_arguments): Likewise.
12604 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12605 int_mode_for_mode instead of mode_for_size.
12606 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12609 Alan Hayward <alan.hayward@arm.com>
12610 David Sherwood <david.sherwood@arm.com>
12612 * machmode.h (int_mode_for_size): New function.
12613 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12614 instead of mode_for_size.
12615 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12617 * combine.c (expand_field_assignment): Use int_mode_for_size
12618 instead of mode_for_size.
12619 (make_extraction): Likewise.
12620 (simplify_shift_const_1): Likewise.
12621 (simplify_comparison): Likewise.
12622 * dojump.c (do_jump): Likewise.
12623 * dwarf2out.c (mem_loc_descriptor): Likewise.
12624 * emit-rtl.c (init_derived_machine_modes): Likewise.
12625 * expmed.c (flip_storage_order): Likewise.
12626 (convert_extracted_bit_field): Likewise.
12627 * expr.c (copy_blkmode_from_reg): Likewise.
12628 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12629 * internal-fn.c (expand_mul_overflow): Likewise.
12630 * lower-subreg.c (simple_move): Likewise.
12631 * optabs-libfuncs.c (init_optabs): Likewise.
12632 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12633 * tree.c (vector_type_mode): Likewise.
12634 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12635 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12636 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12637 * tree-vect-stmts.c (vectorizable_load): Likewise.
12638 (vectorizable_store): Likewise.
12640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12641 Alan Hayward <alan.hayward@arm.com>
12642 David Sherwood <david.sherwood@arm.com>
12644 * coretypes.h (pod_mode): New type.
12645 (scalar_int_mode_pod): New typedef.
12646 * machmode.h (pod_mode): New class.
12647 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12648 * genmodes.c (emit_mode_int_n): Update accordingly.
12649 * lower-subreg.h (target_lower_subreg): Change type to
12650 scalar_int_mode_pod.
12651 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12652 scalar_int_mode_pod.
12654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12655 Alan Hayward <alan.hayward@arm.com>
12656 David Sherwood <david.sherwood@arm.com>
12658 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12659 machine_mode to scalar_int_mode.
12660 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12661 (rs6000_option_override_internal): Remove cast to int.
12662 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12663 machine_mode to scalar_int_mode.
12664 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12665 (rs6000_option_override_internal): Remove cast to int.
12666 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12667 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12669 * config/s390/s390.c (s390_expand_builtin): Likewise.
12670 * coretypes.h (scalar_int_mode): New type.
12671 (opt_scalar_int_mode): New typedef.
12672 * machmode.h (scalar_int_mode): New class.
12673 (scalar_int_mode::includes_p): New function.
12674 (byte_mode): Change type to scalar_int_mode.
12675 (word_mode): Likewise.
12676 (ptr_mode): Likewise.
12677 * emit-rtl.c (byte_mode): Likewise.
12678 (word_mode): Likewise.
12679 (ptr_mode): Likewise.
12680 (init_derived_machine_modes): Update accordingly.
12681 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12682 and MODE_PARTIAL_INT.
12683 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12684 opt_scalar_int_mode.
12686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12687 Alan Hayward <alan.hayward@arm.com>
12688 David Sherwood <david.sherwood@arm.com>
12690 * target.def (libgcc_floating_mode_supported_p): Take a
12692 * doc/tm.texi: Regenerate.
12693 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12695 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12696 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12700 Alan Hayward <alan.hayward@arm.com>
12701 David Sherwood <david.sherwood@arm.com>
12703 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12704 * doc/tm.texi: Regenerate.
12705 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12706 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12707 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12708 * targhooks.h (default_floatn_mode): Likewise.
12709 * targhooks.c (default_floatn_mode): Likewise.
12710 * tree.c (build_common_tree_nodes): Update accordingly.
12712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12713 Alan Hayward <alan.hayward@arm.com>
12714 David Sherwood <david.sherwood@arm.com>
12716 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12717 (mode_iterator::iterate_p): Likewise.
12718 (mode_iterator::get_wider): Likewise.
12719 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12722 Alan Hayward <alan.hayward@arm.com>
12723 David Sherwood <david.sherwood@arm.com>
12725 * coretypes.h (opt_scalar_float_mode): New typedef.
12726 * machmode.h (float_mode_for_size): New function.
12727 * emit-rtl.c (double_mode): Delete.
12728 (init_emit_once): Use float_mode_for_size.
12729 * stor-layout.c (layout_type): Likewise.
12730 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12733 Alan Hayward <alan.hayward@arm.com>
12734 David Sherwood <david.sherwood@arm.com>
12736 * output.h (assemble_real): Take a scalar_float_mode.
12737 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12738 * config/arm/arm.md (consttable_4): Likewise.
12739 (consttable_8): Likewise.
12740 (consttable_16): Likewise.
12741 * config/mips/mips.md (consttable_float): Likewise.
12742 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12743 * varasm.c (assemble_real): Take a scalar_float_mode.
12744 (output_constant_pool_2): Update accordingly.
12745 (output_constant): Likewise.
12747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12748 Alan Hayward <alan.hayward@arm.com>
12749 David Sherwood <david.sherwood@arm.com>
12751 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12752 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12753 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12754 (native_encode_real): Likewise.
12755 (native_interpret_real): Likewise.
12756 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12757 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12760 Alan Hayward <alan.hayward@arm.com>
12761 David Sherwood <david.sherwood@arm.com>
12763 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12764 <scalar_float_mode>. Simplify.
12765 (gen_extend_conv_libfunc): Likewise.
12767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * coretypes.h (scalar_float_mode): New type.
12772 * machmode.h (mode_traits::from_int): Use machine_mode if
12773 USE_ENUM_MODES is defined.
12774 (is_a): New function.
12776 (dyn_cast): Likewise.
12777 (scalar_float_mode): New class.
12778 (scalar_float_mode::includes_p): New function.
12779 (is_float_mode): Likewise.
12780 * gdbhooks.py (MachineModePrinter): New class.
12781 (build_pretty_printer): Use it for scalar_float_mode.
12782 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12783 (format_helper::format_helper): Turn into a template.
12784 * genmodes.c (get_mode_class): New function.
12785 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12786 or machine_mode if none.
12787 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12788 as_a <scalar_float_mode>.
12789 * dwarf2out.c (mem_loc_descriptor): Likewise.
12790 (insert_float): Likewise.
12791 (add_const_value_attribute): Likewise.
12792 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12793 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12794 (expand_unop): Update accordingly.
12795 (expand_abs_nojump): Likewise.
12796 (expand_copysign_absneg): Take a scalar_float_mode.
12797 (expand_copysign_bit): Likewise.
12798 (expand_copysign): Update accordingly.
12800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12801 Alan Hayward <alan.hayward@arm.com>
12802 David Sherwood <david.sherwood@arm.com>
12804 * coretypes.h (opt_mode): New class.
12805 * machmode.h (opt_mode): Likewise.
12806 (opt_mode::else_void): New function.
12807 (opt_mode::require): Likewise.
12808 (opt_mode::exists): Likewise.
12809 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12810 (GET_MODE_2XWIDER_MODE): Likewise.
12811 (mode_iterator::get_wider): Update accordingly.
12812 (mode_iterator::get_2xwider): Likewise.
12813 (mode_iterator::get_known_wider): Likewise, turning into a template.
12814 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12815 forcing a wider mode to exist.
12816 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12817 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12818 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12819 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12820 * lower-subreg.c (init_lower_subreg): Likewise.
12821 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12822 on the final iteration.
12823 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12824 a wider mode exists before asking for a move pattern.
12825 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12826 forcing a wider mode to exist.
12827 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12828 returning false if no such mode exists.
12829 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12830 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12831 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12832 Avoid checking for a MODE_INT if we already know the mode is not a
12834 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12835 forcing a wider mode to exist.
12836 (expmed_mult_highpart_optab): Likewise.
12837 (expmed_mult_highpart): Likewise.
12838 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12840 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12841 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12842 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12843 * internal-fn.c (expand_mul_overflow): Update use of
12844 GET_MODE_2XWIDER_MODE.
12845 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12846 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12847 GET_MODE_WIDER_MODE.
12848 (convert_plusminus_to_widen): Likewise.
12849 * tree-switch-conversion.c (array_value_type): Likewise.
12850 * var-tracking.c (emit_note_insn_var_location): Likewise.
12851 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12852 Return false inside rather than outside the loop if no wider mode
12854 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12855 and GET_MODE_2XWIDER_MODE
12856 (can_compare_p): Use else_void.
12857 * gdbhooks.py (OptMachineModePrinter): New class.
12858 (build_pretty_printer): Use it for opt_mode.
12860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12861 Alan Hayward <alan.hayward@arm.com>
12862 David Sherwood <david.sherwood@arm.com>
12864 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12865 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12868 Alan Hayward <alan.hayward@arm.com>
12869 David Sherwood <david.sherwood@arm.com>
12871 * machmode.h (mode_traits): New structure.
12872 (get_narrowest_mode): New function.
12873 (mode_iterator::start): Likewise.
12874 (mode_iterator::iterate_p): Likewise.
12875 (mode_iterator::get_wider): Likewise.
12876 (mode_iterator::get_known_wider): Likewise.
12877 (mode_iterator::get_2xwider): Likewise.
12878 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12879 (FOR_EACH_MODE): Likewise.
12880 (FOR_EACH_MODE_FROM): Likewise.
12881 (FOR_EACH_MODE_UNTIL): Likewise.
12882 (FOR_EACH_WIDER_MODE): Likewise.
12883 (FOR_EACH_2XWIDER_MODE): Likewise.
12884 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12885 * combine.c (simplify_comparison): Likewise
12886 * config/i386/i386.c (type_natural_mode): Likewise.
12887 * cse.c (cse_insn): Likewise.
12888 * dse.c (find_shift_sequence): Likewise.
12889 * emit-rtl.c (init_derived_machine_modes): Likewise.
12890 (init_emit_once): Likewise.
12891 * explow.c (hard_function_value): Likewise.
12892 * expmed.c (extract_fixed_bit_field_1): Likewise.
12893 (extract_bit_field_1): Likewise.
12894 (expand_divmod): Likewise.
12895 (emit_store_flag_1): Likewise.
12896 * expr.c (init_expr_target): Likewise.
12897 (convert_move): Likewise.
12898 (alignment_for_piecewise_move): Likewise.
12899 (widest_int_mode_for_size): Likewise.
12900 (emit_block_move_via_movmem): Likewise.
12901 (copy_blkmode_to_reg): Likewise.
12902 (set_storage_via_setmem): Likewise.
12903 (compress_float_constant): Likewise.
12904 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12905 * optabs-query.c (get_best_extraction_insn): Likewise.
12906 * optabs.c (expand_binop): Likewise.
12907 (expand_twoval_unop): Likewise.
12908 (expand_twoval_binop): Likewise.
12909 (widen_leading): Likewise.
12910 (widen_bswap): Likewise.
12911 (expand_parity): Likewise.
12912 (expand_unop): Likewise.
12913 (prepare_cmp_insn): Likewise.
12914 (prepare_float_lib_cmp): Likewise.
12915 (expand_float): Likewise.
12916 (expand_fix): Likewise.
12917 (expand_sfix_optab): Likewise.
12918 * postreload.c (move2add_use_add2_insn): Likewise.
12919 * reg-stack.c (reg_to_stack): Likewise.
12920 * reginfo.c (choose_hard_reg_mode): Likewise.
12921 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12922 * stor-layout.c (mode_for_size): Likewise.
12923 (smallest_mode_for_size): Likewise.
12924 (mode_for_vector): Likewise.
12925 (finish_bitfield_representative): Likewise.
12926 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12927 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12928 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12929 * var-tracking.c (prepare_call_arguments): Likewise.
12931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12932 Alan Hayward <alan.hayward@arm.com>
12933 David Sherwood <david.sherwood@arm.com>
12935 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12936 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12937 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12938 * machmode.h (mode_size): Move earlier in file.
12939 (mode_precision): Likewise.
12940 (mode_inner): Likewise.
12941 (mode_nunits): Likewise.
12942 (mode_unit_size): Likewise.
12943 (unit_unit_precision): Likewise.
12944 (mode_wider): Likewise.
12945 (mode_2xwider): Likewise.
12946 (machine_mode): New class.
12947 (mode_to_bytes): New function.
12948 (mode_to_bits): Likewise.
12949 (mode_to_precision): Likewise.
12950 (mode_to_inner): Likewise.
12951 (mode_to_unit_size): Likewise.
12952 (mode_to_unit_precision): Likewise.
12953 (mode_to_nunits): Likewise.
12954 (GET_MODE_SIZE): Use mode_to_bytes.
12955 (GET_MODE_BITSIZE): Use mode_to_bits.
12956 (GET_MODE_PRECISION): Use mode_to_precision.
12957 (GET_MODE_INNER): Use mode_to_inner.
12958 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12959 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12960 (GET_MODE_NUNITS): Use mode_to_nunits.
12961 * system.h (ALWAYS_INLINE): New macro.
12962 * config/powerpcspe/powerpcspe-c.c
12963 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12964 int for arg1_mode and arg2_mode.
12966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12967 Alan Hayward <alan.hayward@arm.com>
12968 David Sherwood <david.sherwood@arm.com>
12970 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12971 Prefix mode names with E_ in case statements.
12972 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12973 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12974 (aarch64_split_simd_move): Likewise.
12975 (aarch64_gen_storewb_pair): Likewise.
12976 (aarch64_gen_loadwb_pair): Likewise.
12977 (aarch64_gen_store_pair): Likewise.
12978 (aarch64_gen_load_pair): Likewise.
12979 (aarch64_get_condition_code_1): Likewise.
12980 (aarch64_constant_pool_reload_icode): Likewise.
12981 (get_rsqrte_type): Likewise.
12982 (get_rsqrts_type): Likewise.
12983 (get_recpe_type): Likewise.
12984 (get_recps_type): Likewise.
12985 (aarch64_gimplify_va_arg_expr): Likewise.
12986 (aarch64_simd_container_mode): Likewise.
12987 (aarch64_emit_load_exclusive): Likewise.
12988 (aarch64_emit_store_exclusive): Likewise.
12989 (aarch64_expand_compare_and_swap): Likewise.
12990 (aarch64_gen_atomic_cas): Likewise.
12991 (aarch64_emit_bic): Likewise.
12992 (aarch64_emit_atomic_swap): Likewise.
12993 (aarch64_emit_atomic_load_op): Likewise.
12994 (aarch64_evpc_trn): Likewise.
12995 (aarch64_evpc_uzp): Likewise.
12996 (aarch64_evpc_zip): Likewise.
12997 (aarch64_evpc_ext): Likewise.
12998 (aarch64_evpc_rev): Likewise.
12999 (aarch64_evpc_dup): Likewise.
13000 (aarch64_gen_ccmp_first): Likewise.
13001 (aarch64_gen_ccmp_next): Likewise.
13002 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13003 (alpha_emit_xfloating_libcall): Likewise.
13004 (emit_insxl): Likewise.
13005 (alpha_arg_type): Likewise.
13006 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13007 (arc_preferred_simd_mode): Likewise.
13008 (arc_secondary_reload): Likewise.
13009 (get_arc_condition_code): Likewise.
13010 (arc_print_operand): Likewise.
13011 (arc_legitimate_constant_p): Likewise.
13012 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13013 * config/arc/arc.md (casesi_load): Likewise.
13014 (casesi_compact_jump): Likewise.
13015 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13016 (cc_use_register): Likewise.
13017 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13018 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13019 (arm_init_iwmmxt_builtins): Likewise.
13020 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13021 (neon_expand_vector_init): Likewise.
13022 (arm_attr_length_move_neon): Likewise.
13023 (maybe_get_arm_condition_code): Likewise.
13024 (arm_emit_vector_const): Likewise.
13025 (arm_preferred_simd_mode): Likewise.
13026 (arm_output_iwmmxt_tinsr): Likewise.
13027 (thumb1_output_casesi): Likewise.
13028 (thumb2_output_casesi): Likewise.
13029 (arm_emit_load_exclusive): Likewise.
13030 (arm_emit_store_exclusive): Likewise.
13031 (arm_expand_compare_and_swap): Likewise.
13032 (arm_evpc_neon_vuzp): Likewise.
13033 (arm_evpc_neon_vzip): Likewise.
13034 (arm_evpc_neon_vrev): Likewise.
13035 (arm_evpc_neon_vtrn): Likewise.
13036 (arm_evpc_neon_vext): Likewise.
13037 (arm_validize_comparison): Likewise.
13038 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13039 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13040 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13041 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13042 (c6x_preferred_simd_mode): Likewise.
13043 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13044 (epiphany_rtx_costs): Likewise.
13045 * config/epiphany/predicates.md (proper_comparison_operator):
13047 * config/frv/frv.c (condexec_memory_operand): Likewise.
13048 (frv_emit_move): Likewise.
13049 (output_move_single): Likewise.
13050 (output_condmove_single): Likewise.
13051 (frv_hard_regno_mode_ok): Likewise.
13052 (frv_matching_accg_mode): Likewise.
13053 * config/h8300/h8300.c (split_adds_subs): Likewise.
13054 (h8300_rtx_costs): Likewise.
13055 (h8300_print_operand): Likewise.
13056 (compute_mov_length): Likewise.
13057 (output_logical_op): Likewise.
13058 (compute_logical_op_length): Likewise.
13059 (compute_logical_op_cc): Likewise.
13060 (h8300_shift_needs_scratch_p): Likewise.
13061 (output_a_shift): Likewise.
13062 (compute_a_shift_length): Likewise.
13063 (compute_a_shift_cc): Likewise.
13064 (expand_a_rotate): Likewise.
13065 (output_a_rotate): Likewise.
13066 * config/i386/i386.c (classify_argument): Likewise.
13067 (function_arg_advance_32): Likewise.
13068 (function_arg_32): Likewise.
13069 (function_arg_64): Likewise.
13070 (function_value_64): Likewise.
13071 (ix86_gimplify_va_arg): Likewise.
13072 (ix86_legitimate_constant_p): Likewise.
13073 (put_condition_code): Likewise.
13074 (split_double_mode): Likewise.
13075 (ix86_avx256_split_vector_move_misalign): Likewise.
13076 (ix86_expand_vector_logical_operator): Likewise.
13077 (ix86_split_idivmod): Likewise.
13078 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13079 (ix86_build_const_vector): Likewise.
13080 (ix86_build_signbit_mask): Likewise.
13081 (ix86_match_ccmode): Likewise.
13082 (ix86_cc_modes_compatible): Likewise.
13083 (ix86_expand_branch): Likewise.
13084 (ix86_expand_sse_cmp): Likewise.
13085 (ix86_expand_sse_movcc): Likewise.
13086 (ix86_expand_int_sse_cmp): Likewise.
13087 (ix86_expand_vec_perm_vpermi2): Likewise.
13088 (ix86_expand_vec_perm): Likewise.
13089 (ix86_expand_sse_unpack): Likewise.
13090 (ix86_expand_int_addcc): Likewise.
13091 (ix86_split_to_parts): Likewise.
13092 (ix86_vectorize_builtin_gather): Likewise.
13093 (ix86_vectorize_builtin_scatter): Likewise.
13094 (avx_vpermilp_parallel): Likewise.
13095 (inline_memory_move_cost): Likewise.
13096 (ix86_tieable_integer_mode_p): Likewise.
13097 (x86_maybe_negate_const_int): Likewise.
13098 (ix86_expand_vector_init_duplicate): Likewise.
13099 (ix86_expand_vector_init_one_nonzero): Likewise.
13100 (ix86_expand_vector_init_one_var): Likewise.
13101 (ix86_expand_vector_init_concat): Likewise.
13102 (ix86_expand_vector_init_interleave): Likewise.
13103 (ix86_expand_vector_init_general): Likewise.
13104 (ix86_expand_vector_set): Likewise.
13105 (ix86_expand_vector_extract): Likewise.
13106 (emit_reduc_half): Likewise.
13107 (ix86_emit_i387_round): Likewise.
13108 (ix86_mangle_type): Likewise.
13109 (ix86_expand_round_sse4): Likewise.
13110 (expand_vec_perm_blend): Likewise.
13111 (canonicalize_vector_int_perm): Likewise.
13112 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13113 (expand_vec_perm_1): Likewise.
13114 (expand_vec_perm_interleave3): Likewise.
13115 (expand_vec_perm_even_odd_pack): Likewise.
13116 (expand_vec_perm_even_odd_1): Likewise.
13117 (expand_vec_perm_broadcast_1): Likewise.
13118 (ix86_vectorize_vec_perm_const_ok): Likewise.
13119 (ix86_expand_vecop_qihi): Likewise.
13120 (ix86_expand_mul_widen_hilo): Likewise.
13121 (ix86_expand_sse2_abs): Likewise.
13122 (ix86_expand_pextr): Likewise.
13123 (ix86_expand_pinsr): Likewise.
13124 (ix86_preferred_simd_mode): Likewise.
13125 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13126 * config/i386/sse.md (*andnot<mode>3): Likewise.
13127 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13128 (*<code><mode>3): Likewise.
13129 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13130 (ia64_expand_atomic_op): Likewise.
13131 (ia64_arg_type): Likewise.
13132 (ia64_mode_to_int): Likewise.
13133 (ia64_scalar_mode_supported_p): Likewise.
13134 (ia64_vector_mode_supported_p): Likewise.
13135 (expand_vec_perm_broadcast): Likewise.
13136 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13137 (iq2000_function_arg_advance): Likewise.
13138 (iq2000_function_arg): Likewise.
13139 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13140 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13141 (m68k_libcall_value): Likewise.
13142 (m68k_function_value): Likewise.
13143 (sched_attr_op_type): Likewise.
13144 * config/mcore/mcore.c (mcore_output_move): Likewise.
13145 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13147 (microblaze_function_arg): Likewise.
13148 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13149 (mips_print_operand): Likewise.
13150 (mips_mode_ok_for_mov_fmt_p): Likewise.
13151 (mips_vector_mode_supported_p): Likewise.
13152 (mips_preferred_simd_mode): Likewise.
13153 (mips_expand_vpc_loongson_even_odd): Likewise.
13154 (mips_expand_vec_unpack): Likewise.
13155 (mips_expand_vi_broadcast): Likewise.
13156 (mips_expand_vector_init): Likewise.
13157 (mips_expand_vec_reduc): Likewise.
13158 (mips_expand_msa_cmp): Likewise.
13159 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13160 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13161 (cc_flags_for_mode): Likewise.
13162 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13163 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13164 (nds32_output_casesi_pc_relative): Likewise.
13165 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13166 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13167 (nvptx_gen_unpack): Likewise.
13168 (nvptx_gen_pack): Likewise.
13169 (nvptx_gen_shuffle): Likewise.
13170 (nvptx_gen_wcast): Likewise.
13171 (nvptx_preferred_simd_mode): Likewise.
13172 * config/pa/pa.c (pa_secondary_reload): Likewise.
13173 * config/pa/predicates.md (base14_operand): Likewise.
13174 * config/powerpcspe/powerpcspe-c.c
13175 (altivec_resolve_overloaded_builtin): Likewise.
13176 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13178 (rs6000_preferred_simd_mode): Likewise.
13179 (output_vec_const_move): Likewise.
13180 (rs6000_expand_vector_extract): Likewise.
13181 (rs6000_split_vec_extract_var): Likewise.
13182 (reg_offset_addressing_ok_p): Likewise.
13183 (rs6000_legitimate_offset_address_p): Likewise.
13184 (rs6000_legitimize_address): Likewise.
13185 (rs6000_emit_set_const): Likewise.
13186 (rs6000_const_vec): Likewise.
13187 (rs6000_emit_move): Likewise.
13188 (spe_build_register_parallel): Likewise.
13189 (rs6000_darwin64_record_arg_recurse): Likewise.
13190 (swap_selector_for_mode): Likewise.
13191 (spe_init_builtins): Likewise.
13192 (paired_init_builtins): Likewise.
13193 (altivec_init_builtins): Likewise.
13194 (do_load_for_compare): Likewise.
13195 (rs6000_generate_compare): Likewise.
13196 (rs6000_expand_float128_convert): Likewise.
13197 (emit_load_locked): Likewise.
13198 (emit_store_conditional): Likewise.
13199 (rs6000_output_function_epilogue): Likewise.
13200 (rs6000_handle_altivec_attribute): Likewise.
13201 (rs6000_function_value): Likewise.
13202 (emit_fusion_gpr_load): Likewise.
13203 (emit_fusion_p9_load): Likewise.
13204 (emit_fusion_p9_store): Likewise.
13205 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13206 (fusion_gpr_mem_load): Likewise.
13207 (fusion_addis_mem_combo_load): Likewise.
13208 (fusion_addis_mem_combo_store): Likewise.
13209 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13210 (fusion_gpr_mem_load): Likewise.
13211 (fusion_addis_mem_combo_load): Likewise.
13212 (fusion_addis_mem_combo_store): Likewise.
13213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13215 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13216 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13217 (rs6000_preferred_simd_mode): Likewise.
13218 (output_vec_const_move): Likewise.
13219 (rs6000_expand_vector_extract): Likewise.
13220 (rs6000_split_vec_extract_var): Likewise.
13221 (reg_offset_addressing_ok_p): Likewise.
13222 (rs6000_legitimate_offset_address_p): Likewise.
13223 (rs6000_legitimize_address): Likewise.
13224 (rs6000_emit_set_const): Likewise.
13225 (rs6000_const_vec): Likewise.
13226 (rs6000_emit_move): Likewise.
13227 (rs6000_darwin64_record_arg_recurse): Likewise.
13228 (swap_selector_for_mode): Likewise.
13229 (paired_init_builtins): Likewise.
13230 (altivec_init_builtins): Likewise.
13231 (rs6000_expand_float128_convert): Likewise.
13232 (emit_load_locked): Likewise.
13233 (emit_store_conditional): Likewise.
13234 (rs6000_output_function_epilogue): Likewise.
13235 (rs6000_handle_altivec_attribute): Likewise.
13236 (rs6000_function_value): Likewise.
13237 (emit_fusion_gpr_load): Likewise.
13238 (emit_fusion_p9_load): Likewise.
13239 (emit_fusion_p9_store): Likewise.
13240 * config/rx/rx.c (rx_gen_move_template): Likewise.
13241 (flags_from_mode): Likewise.
13242 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13243 (s390_slb_comparison): Likewise.
13244 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13245 (s390_vector_mode_supported_p): Likewise.
13246 (s390_cc_modes_compatible): Likewise.
13247 (s390_match_ccmode_set): Likewise.
13248 (s390_canonicalize_comparison): Likewise.
13249 (s390_emit_compare_and_swap): Likewise.
13250 (s390_branch_condition_mask): Likewise.
13251 (s390_rtx_costs): Likewise.
13252 (s390_secondary_reload): Likewise.
13253 (__SECONDARY_RELOAD_CASE): Likewise.
13254 (s390_expand_cs): Likewise.
13255 (s390_preferred_simd_mode): Likewise.
13256 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13257 * config/sh/sh.c (sh_print_operand): Likewise.
13258 (dump_table): Likewise.
13259 (sh_secondary_reload): Likewise.
13260 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13261 * config/sh/sh.md (casesi_worker_1): Likewise.
13262 (casesi_worker_2): Likewise.
13263 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13264 (fcc_comparison_operator): Likewise.
13265 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13266 (emit_soft_tfmode_cvt): Likewise.
13267 (sparc_preferred_simd_mode): Likewise.
13268 (output_cbranch): Likewise.
13269 (sparc_print_operand): Likewise.
13270 (sparc_expand_vec_perm_bmask): Likewise.
13271 (vector_init_bshuffle): Likewise.
13272 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13273 (spu_vector_mode_supported_p): Likewise.
13274 (spu_expand_insv): Likewise.
13275 (spu_emit_branch_or_set): Likewise.
13276 (spu_handle_vector_attribute): Likewise.
13277 (spu_builtin_splats): Likewise.
13278 (spu_builtin_extract): Likewise.
13279 (spu_builtin_promote): Likewise.
13280 (spu_expand_sign_extend): Likewise.
13281 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13282 (tilegx_simd_int): Likewise.
13283 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13284 (tilepro_simd_int): Likewise.
13285 * config/v850/v850.c (const_double_split): Likewise.
13286 (v850_print_operand): Likewise.
13287 (ep_memory_offset): Likewise.
13288 * config/vax/vax.c (vax_rtx_costs): Likewise.
13289 (vax_output_int_move): Likewise.
13290 (vax_output_int_add): Likewise.
13291 (vax_output_int_subtract): Likewise.
13292 * config/visium/predicates.md (visium_branch_operator): Likewise.
13293 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13294 (visium_print_operand_address): Likewise.
13295 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13296 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13297 (xtensa_expand_conditional_branch): Likewise.
13298 (xtensa_copy_incoming_a7): Likewise.
13299 (xtensa_output_literal): Likewise.
13300 * dfp.c (decimal_real_maxval): Likewise.
13301 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13304 Alan Hayward <alan.hayward@arm.com>
13305 David Sherwood <david.sherwood@arm.com>
13307 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13308 (mode_nunits_inline): Likewise.
13309 (mode_inner_inline): Likewise.
13310 (mode_unit_size_inline): Likewise.
13311 (mode_unit_precision_inline): Likewise.
13312 (emit_insn_modes_h): Likewise. Also emit a #define of the
13314 (emit_mode_wider): Add an E_ prefix to mode names.
13315 (emit_mode_complex): Likewise.
13316 (emit_mode_inner): Likewise.
13317 (emit_mode_adjustments): Likewise.
13318 (emit_mode_int_n): Likewise.
13319 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13320 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13321 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13322 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13323 (CRC32_BUILTIN, ENTRY): Likewise.
13324 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13325 (aarch64_pop_regs): Likewise.
13326 (aarch64_process_components): Likewise.
13327 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13328 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13329 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13330 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13331 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13332 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13334 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13335 (pcmpistr): Likewise.
13336 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13337 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13338 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13340 * config/rl78/rl78.c (mduc_regs): Likewise.
13341 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13342 (htm_expand_builtin): Likewise.
13343 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13344 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13345 * config/xtensa/xtensa.c (print_operand): Likewise.
13346 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13347 (NUM_MODE_VECTOR_INT): Likewise.
13348 * genoutput.c (null_operand): Likewise.
13349 (output_operand_data): Likewise.
13350 * genrecog.c (print_parameter_value): Likewise.
13351 * lra.c (debug_operand_data): Likewise.
13353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13355 * dwarf2out.c (scompare_loc_descriptor_wide)
13356 (scompare_loc_descriptor_narrow): New functions, split out from...
13357 (scompare_loc_descriptor): ...here.
13358 * expmed.c (emit_store_flag_int): New function, split out from...
13359 (emit_store_flag): ...here.
13361 2017-08-30 Richard Biener <rguenther@suse.de>
13363 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13364 (dwarf2out_early_finish): Move setting of AT_pubnames from
13365 early debug output to early finish.
13367 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13369 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13370 and -mdata-region to the assembler.
13372 2017-08-30 Richard Biener <rguenther@suse.de>
13374 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13376 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13377 (dwarf2out_early_global_decl): Only generate a DIE for the
13378 abstract origin if it doesn't already exist or is a declaration DIE.
13379 (resolve_addr): Do not add the linkage name twice when
13380 generating a stub DIE for the DW_TAG_GNU_call_site target.
13382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13384 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13385 Use machine_mode rather than int for arg1_mode.
13387 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13390 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13391 that the second argument of the built-in functions to unpack
13392 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13393 switch statement instead a lot of if statements.
13394 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13395 Allow 64-bit values to be in Altivec registers as well as
13396 traditional floating point registers.
13397 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13399 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13401 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13402 MAX_REGS_PER_ADDRESS == 1.
13404 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13406 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13407 * config/i386/i386.c (ix86_option_override_internal): Simplify
13408 setting of opts->x_flag_entry.
13410 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13411 Jakub Jelinek <jakub@redhat.com>
13412 Richard Biener <rguenther@suse.de>
13414 PR tree-optimization/81503
13415 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13416 folded constant fits in the target type; reorder tests for clarity.
13418 2017-08-29 Martin Liska <mliska@suse.cz>
13420 * passes.def: Include pass_lower_switch.
13421 * stmt.c (dump_case_nodes): Remove and move to
13422 tree-switch-conversion.
13423 (case_values_threshold): Likewise.
13424 (expand_switch_as_decision_tree_p): Likewise.
13425 (emit_case_decision_tree): Likewise.
13426 (expand_case): Likewise.
13427 (balance_case_nodes): Likewise.
13428 (node_has_low_bound): Likewise.
13429 (node_has_high_bound): Likewise.
13430 (node_is_bounded): Likewise.
13431 (emit_case_nodes): Likewise.
13432 (struct simple_case_node): New struct.
13433 (add_case_node): Remove.
13434 (emit_case_dispatch_table): Use vector instead of case_list.
13435 (reset_out_edges_aux): Remove.
13436 (compute_cases_per_edge): Likewise.
13437 (expand_case): Build list of simple_case_node.
13438 (expand_sjlj_dispatch_table): Use it.
13439 * tree-switch-conversion.c (struct case_node): Moved from
13440 stmt.c and adjusted.
13441 (emit_case_nodes): Likewise.
13442 (node_has_low_bound): Likewise.
13443 (node_has_high_bound): Likewise.
13444 (node_is_bounded): Likewise.
13445 (case_values_threshold): Likewise.
13446 (reset_out_edges_aux): Likewise.
13447 (compute_cases_per_edge): Likewise.
13448 (add_case_node): Likewise.
13449 (dump_case_nodes): Likewise.
13450 (balance_case_nodes): Likewise.
13451 (expand_switch_as_decision_tree_p): Likewise.
13452 (emit_jump): Likewise.
13453 (emit_case_decision_tree): Likewise.
13454 (try_switch_expansion): Likewise.
13455 (do_jump_if_equal): Likewise.
13456 (emit_cmp_and_jump_insns): Likewise.
13457 (fix_phi_operands_for_edge): New function.
13458 (record_phi_operand_mapping): Likewise.
13459 (class pass_lower_switch): New pass.
13460 (pass_lower_switch::execute): New function.
13461 (make_pass_lower_switch): Likewise.
13462 (conditional_probability):
13463 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13464 * tree-pass.h: Add make_pass_lower_switch.
13466 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13469 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13472 2017-08-29 Richard Biener <rguenther@suse.de>
13474 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13475 we do not add a DW_AT_inline attribute twice.
13476 (gen_subprogram_die): Remove code setting DW_AT_inline on
13477 DECL_ABSTRACT_P nodes.
13479 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13481 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13482 calls to internal functions.
13483 (gimplify_modify_expr): Likewise.
13484 * tree-call-cdce.c (use_internal_fn): Likewise.
13485 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13486 (convert_to_divmod): Set the nothrow flag.
13487 * tree-if-conv.c (predicate_mem_writes): Likewise.
13488 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13489 (vectorizable_call): Likewise.
13490 (vectorizable_store): Likewise.
13491 (vectorizable_load): Likewise.
13492 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13493 (vect_recog_mask_conversion_pattern): Likewise.
13495 2017-08-29 Martin Liska <mliska@suse.cz>
13498 * gcc.c (driver_handle_option): Add new argument.
13499 * opts-common.c (handle_option): Pass
13500 target_option_override_hook.
13501 * opts-global.c (lang_handle_option): Add new option.
13502 (set_default_handlers): Add new argument.
13503 (decode_options): Likewise.
13504 * opts.c (target_handle_option): Likewise.
13505 (common_handle_option): Call target_option_override_hook.
13506 * opts.h (struct cl_option_handler_func): Add hook for
13507 target option override.
13508 (struct cl_option_handlers): Likewise.
13509 (set_default_handlers): Add new argument.
13510 (decode_options): Likewise.
13511 (common_handle_option): Likewise.
13512 (target_handle_option): Likewise.
13513 * toplev.c (toplev::main): Pass targetm.target_option.override
13516 2017-08-29 Richard Biener <rguenther@suse.de>
13517 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13519 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13520 life to the active subtree.
13522 2017-08-28 Jeff Law <law@redhat.com>
13524 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13525 derive_equivalences.
13526 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13527 Code moved into....
13528 (edge_info::derive_equivalences): New private member function
13530 * tree-ssa-dom.c (class edge_info): Changed from a struct
13531 to a class. Add ctor/dtor, methods and data members.
13532 (edge_info::edge_info): Renamed from allocate_edge_info.
13533 Initialize additional members.
13534 (edge_info::~edge_info): New.
13535 (free_dom_edge_info): Delete the edge info.
13536 (record_edge_info): Use new class & associated member functions.
13537 Tighten forms for testing for edge equivalences.
13538 (record_temporary_equivalences): Iterate over the simple
13539 equivalences rather than assuming there's only one per edge.
13540 (cprop_into_successor_phis): Iterate over the simple
13541 equivalences rather than assuming there's only one per edge.
13542 (optimize_stmt): Use operand_equal_p rather than pointer
13543 equality for mini-DSE code.
13545 2017-08-28 Nathan Sidwell <nathan@acm.org>
13547 * gcc.c (execute): Fold SIGPIPE handling into switch
13548 statement. Adjust internal error message.
13550 2017-08-28 Richard Biener <rguenther@suse.de>
13553 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13554 Do nothing for removed DIEs.
13556 2017-08-28 Richard Biener <rguenther@suse.de>
13558 PR tree-optimization/81977
13559 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13562 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13565 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13566 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13567 using targetm.gen_mem_thread_fence.
13569 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13572 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13573 predicate to register_operand. Reorder operands.
13574 (*btr<mode>): Ditto.
13575 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13576 (*btr<mode>_mask): Ditto.
13578 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13580 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13581 * config/rs6000/xmmintrin.h: New file.
13582 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13584 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13587 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13588 parameter and_insn and return it.
13589 (recombine_lvx_pattern): Insert a copy to ensure availability of
13590 the base register of the copied masking operation at the point of
13591 the instruction replacement.
13592 (recombine_stvx_pattern): Likewise.
13594 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13596 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13597 undocumented switches.
13598 (-mpower9-dform-vector): Likewise.
13599 (-mpower9-dform): Likewise.
13600 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13601 comments to delete references to -mpower9-dform* switches.
13602 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13603 Delete reference to -mpower9-dform* switches, test for
13604 -mpower9-vector instead.
13605 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13606 (OTHER_P9_VECTOR_MASKS): Likewise.
13607 (POWERPC_MASKS): Likewise.
13608 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13609 tests against -mpower9-dform* to -mpower9-vector. Delete code
13610 that checked for -mpower9-dform* consistancy with other options.
13611 Add test for -mpower9-misc to enable other power9 switches.
13612 (rs6000_init_hard_regno_mode_ok): Likewise.
13613 (rs6000_option_override_internal): Likewise.
13614 (rs6000_emit_prologue): Likewise.
13615 (rs6000_emit_epilogue): Likewise.
13616 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13617 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13618 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13620 (emit_fusion_p9_store): Likewise.
13621 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13622 resetting these macros if the assembler does not support ISA 3.0
13624 (TARGET_P9_DFORM_VECTOR): Likewise.
13625 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13626 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13628 2017-08-25 Alan Modra <amodra@gmail.com>
13630 PR rtl-optimization/81747
13631 * cse.c (cse_extended_basic_block): Don't attempt to record
13632 equivalences for degenerate conditional branches that branch
13633 to their fall-through.
13635 2017-08-24 Martin Sebor <msebor@redhat.com>
13637 PR middle-end/81908
13638 * gimple-fold.c (size_must_be_zero_p): New function.
13639 (gimple_fold_builtin_memory_op): Call it.
13641 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13643 * config/rs6000/mm_malloc.h: New file.
13645 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13647 PR tree-optimization/81913
13648 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13649 analysis when either IVs in condition can wrap.
13651 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13653 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13654 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13656 2017-08-24 Richard Biener <rguenther@suse.de>
13659 * targhooks.c (default_target_can_inline_p): Properly
13660 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13661 is present and always compare.
13662 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13663 infer -mfpmath=sse from TARGET_SSE_P.
13664 (ix86_can_inline_p): Properly use target_option_default_node when
13665 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13667 2017-08-24 Richard Biener <rguenther@suse.de>
13670 * dwarf2out.c (output_die): Handle flag_generate_offload like
13672 (output_comp_unit): Likewise.
13673 (gen_array_type_die): Likewise.
13674 (dwarf2out_early_finish): Likewise.
13675 (note_variable_value_in_expr): Likewise.
13676 (dwarf2out_finish): Likewise. Adjust assert.
13677 * cgraphunit.c (symbol_table::compile): Move setting of
13678 flag_generate_offload earlier ...
13679 (symbol_table::finalize_compilation_unit): ... here, before
13680 early debug finalization.
13682 2017-08-24 Richard Biener <rguenther@suse.de>
13684 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13685 and ipa-fnsummary.h.
13686 (ix86_can_inline_p): When ix86_fpmath flags do not match
13687 check whether the callee uses FP math at all.
13689 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13691 PR middle-end/81931
13692 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13693 instead of TYPE_PRECISION.
13695 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13696 Alan Hayward <alan.hayward@arm.com>
13697 David Sherwood <david.sherwood@arm.com>
13699 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13701 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13703 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13705 (ix86_frame::stack_realign_allocate): New field.
13706 (struct machine_frame_state): Modify comments.
13707 (machine_frame_state::sp_realigned_fp_end): New field.
13708 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13709 layout calculation.
13710 (sp_valid_at): Add assertion to assure no attempt to access invalid
13711 offset of a realigned stack.
13712 (fp_valid_at): Likewise.
13713 (choose_baseaddr): Modify comments.
13714 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13715 ix86_expand_prologue.
13716 (ix86_expand_prologue): Modify stack realignment and allocation.
13717 (ix86_expand_epilogue): Modify comments.
13718 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13719 avx2_runtime, avx512f, and avx512f_runtime.
13721 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13723 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13724 (mstackrealign): Do not init to -1.
13725 * config/i386/i386.c (ix86_option_override_internal):
13726 Check opts_set, not opts when setting default value of
13727 opts->x_ix86_force_align_arg_pointer.
13729 2017-08-23 Richard Biener <rguenther@suse.de>
13731 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13732 lang_hooks.decl_printable_name.
13733 * print-rtl-function.c (print_rtx_function): Likewise.
13734 * tree-pretty-print.c (dump_function_header): Likewise.
13736 2017-08-23 Richard Biener <rguenther@suse.de>
13739 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13740 -g0 at compile-time.
13742 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13744 PR middle-end/19706
13745 * doc/sourcebuild.texi (Other hardware attributes):
13748 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13750 PR middle-end/19706
13751 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13752 Fix single-use check.
13754 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13756 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13758 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13760 * doc/install.texi: Modify to add more details on running selected
13763 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13765 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13766 is combined with -mabi=ms.
13767 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13770 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13772 PR tree-optimization/81488
13773 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13774 and cached_basis fields.
13775 (MAX_SPREAD): New constant.
13776 (alloc_cand_and_find_basis): Initialize new fields.
13777 (clear_visited): New function.
13778 (create_phi_basis_1): Rename from create_phi_basis, set visited
13779 and cached_basis fields.
13780 (create_phi_basis): New wrapper function.
13781 (phi_add_costs_1): Rename from phi_add_costs, add spread
13782 parameter, set visited field, short-circuit when limits reached.
13783 (phi_add_costs): New wrapper function.
13784 (record_phi_increments_1): Rename from record_phi_increments, set
13786 (record_phi_increments): New wrapper function.
13787 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13788 (phi_incr_cost): New wrapper function.
13789 (all_phi_incrs_profitable_1): Rename from
13790 all_phi_incrs_profitable, set visited field.
13791 (all_phi_incrs_profitable): New wrapper function.
13793 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13794 Alan Hayward <alan.hayward@arm.com>
13795 David Sherwood <david.sherwood@arm.com>
13797 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13798 that takes the outer and inner modes.
13799 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13800 comparison as the canonical test for a paradoxical subreg.
13801 * combine.c (simplify_set): Use paradoxical_subreg_p.
13802 (make_extraction): Likewise.
13803 (force_to_mode): Likewise.
13804 (rtx_equal_for_field_assignment_p): Likewise.
13805 (gen_lowpart_for_combine): Likewise.
13806 (simplify_comparison): Likewise.
13807 * cse.c (equiv_constant): Likewise.
13808 * expmed.c (store_bit_field_1): Likewise.
13809 * final.c (alter_subreg): Likewise.
13810 * fwprop.c (propagate_rtx): Likewise.
13811 (forward_propagate_subreg): Likewise.
13812 * ira-conflicts.c (ira_build_conflicts): Likewise.
13813 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13814 * lra-constraints.c (curr_insn_transform): Likewise.
13815 (split_reg): Likewise.
13816 * lra-eliminations.c (move_plus_up): Likewise.
13817 (lra_eliminate_regs_1): Likewise.
13818 * recog.c (general_operand): Likewise.
13819 * ree.c (combine_reaching_defs): Likewise.
13820 * reload.c (push_reload): Likewise.
13821 (find_reloads): Likewise.
13822 * reload1.c (elimination_effects): Likewise.
13823 (compute_reload_subreg_offset): Likewise.
13824 (choose_reload_regs): Likewise.
13825 * rtlanal.c (subreg_lsb_1): Likewise.
13826 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13827 (simplify_subreg): Likewise.
13828 * var-tracking.c (track_loc_p): Likewise.
13829 * emit-rtl.c (byte_lowpart_offset): Likewise.
13830 (paradoxical_subreg_p): Delete out-of-line definition.
13832 2017-08-22 Jeff Law <law@redhat.com>
13834 PR tree-optimization/81741
13835 PR tree-optimization/71947
13836 * tree-ssa-dom.c: Include tree-inline.h.
13837 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13838 equivalences if one is more expensive to compute than the other.
13839 * tree-ssa-scopedtables.h (class const_or_copies): Make
13840 record_const_or_copy_raw method private.
13841 (class avail_exprs_stack): New method simplify_binary_operation.
13842 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13843 avail_exprs_stack::simplify_binary_operation as needed.
13844 (avail_exprs_stack::simplify_binary_operation): New function.
13846 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13848 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13849 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13850 (DOT_SYMBOLS): Likewise.
13851 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13852 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13853 (RS6000_ABI_NAME): Likewise.
13854 (TARGET_CMODEL): Likewise.
13855 (TOC_SECTION_ASM_OP): Likewise.
13856 (SET_CMODEL): New macro.
13857 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13859 2017-08-22 Richard Biener <rguenther@suse.de>
13861 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13862 to allow for free-lang-data replacements similar to verify_type_variant.
13864 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13865 Alan Hayward <alan.hayward@arm.com>
13866 David Sherwood <david.sherwood@arm.com>
13868 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13869 VOIDmode for the LABEL_REF.
13871 2017-08-22 Richard Biener <rguenther@suse.de>
13873 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13875 2017-08-22 Richard Biener <rguenther@suse.de>
13877 * common.opt (feliminate-dwarf2-dups): Ignore.
13878 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13879 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13880 same_die_p_wrap, compute_section_prefix,
13881 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13882 (comdat_symbol_id, comdat_symbol_number): Likewise.
13883 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13885 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13886 (output_die): Mark unreachable path unreachable.
13887 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13888 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13889 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13890 (dwarf2out_early_finish): Likewise.
13892 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13894 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13896 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13899 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13900 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13901 (avr_attribute_table) <io, io_low, address>: Initialize
13902 .decl_required with true.
13904 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13906 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13907 undocumented debugging options.
13908 (-mvsx-scalar-double): Likewise.
13909 (-mallow-df-permute): Likewise.
13910 (-mvectorize-builtins): Likewise.
13911 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13912 (rs6000_builtin_vectorized_function): Likewise.
13913 (rs6000_builtin_md_vectorized_function): Likewise.
13914 (rs6000_opt_vars): Likewise.
13916 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13919 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13920 (*btrq_imm): Rename from *btrq.
13921 (*btcq_imm): Rename from *btcq.
13922 (btsc): New code attribute.
13923 (*<btsc><mode>): New insn pattern.
13924 (*btr<mode>): Ditto.
13925 (*<btsc><mode>_mask): New insn_and_split pattern.
13926 (*btr<mode>_mask): Ditto.
13928 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13929 Alan Hayward <alan.hayward@arm.com>
13930 David Sherwood <david.sherwood@arm.com>
13932 * function.c (pad_below): Simplify padding calculation.
13934 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13935 Alan Hayward <alan.hayward@arm.com>
13936 David Sherwood <david.sherwood@arm.com>
13938 * target.def (function_prologue): Remove frame size argument.
13939 (function_epilogue): Likewise.
13940 * doc/tm.texi: Regenerate.
13941 * final.c (final_start_function): Update call to function_prologue.
13942 (final_end_function): Update call to function_epilogue.
13943 (default_function_pro_epilogue): Remove frame size argument.
13944 * output.h (default_function_pro_epilogue): Likewise.
13945 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13946 (arm_output_function_prologue): Likewise.
13947 * config/frv/frv.c (frv_function_prologue): Likewise.
13948 (frv_function_epilogue): Likewise.
13949 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13950 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13951 (ia64_output_function_epilogue): Likewise.
13952 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13953 (m32r_output_function_epilogue): Likewise.
13954 * config/microblaze/microblaze.c (microblaze_function_prologue)
13955 (microblaze_function_epilogue): Likewise.
13956 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13957 (mips_output_function_epilogue): Likewise.
13958 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13959 (mmix_target_asm_function_epilogue): Likewise.
13960 * config/msp430/msp430.c (msp430_start_function): Likewise.
13961 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13962 (nds32_asm_function_epilogue): Likewise.
13963 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13964 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13965 (pa_output_function_epilogue): Likewise.
13966 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13967 (rs6000_output_function_epilogue): Likewise.
13968 * config/rl78/rl78.c (rl78_start_function): Likewise.
13969 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13970 (rs6000_output_function_epilogue): Likewise.
13971 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13972 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13973 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13974 (sparc_asm_function_epilogue): Likewise.
13976 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13978 * tree.h (type_has_mode_precision_p): New function.
13979 * convert.c (convert_to_integer_1): Use it.
13980 * expr.c (expand_expr_real_2): Likewise.
13981 (expand_expr_real_1): Likewise.
13982 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13983 * match.pd: Likewise.
13984 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13985 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13986 * tree-tailcall.c (process_assignment): Likewise.
13987 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13988 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13989 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13990 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13991 (vectorizable_assignment): Likewise.
13992 (vectorizable_shift): Likewise.
13993 (vectorizable_operation): Likewise.
13994 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13996 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13998 * match.pd: Add pow (C, x) simplification.
14000 2017-08-21 Richard Biener <rguenther@suse.de>
14002 PR tree-optimization/81900
14003 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14004 for blocks with abnormal predecessors.
14005 (compute_antic): Do not set visited flag prematurely.
14007 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14010 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14012 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14014 * stor-layout.h (vector_type_mode): Move to...
14015 * tree.h (vector_type_mode): ...here.
14016 * stor-layout.c (vector_type_mode): Move to...
14017 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14019 2017-08-21 Richard Biener <rguenther@suse.de>
14021 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14022 register_external_die hooks.
14023 (debug_false_tree_charstarstar_uhwistar): Declare.
14024 (debug_nothing_tree_charstar_uhwi): Likewise.
14025 * debug.c (do_nothing_debug_hooks): Adjust.
14026 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14027 (debug_nothing_tree_charstar_uhwi): Likewise.
14028 * dbxout.c (dbx_debug_hooks): Adjust.
14029 (xcoff_debug_hooks): Likewise.
14030 * sdbout.c (sdb_debug_hooks): Likewise.
14031 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14032 * dwarf2out.c (macinfo_label_base): New global.
14033 (dwarf2out_register_external_die): New function for the
14034 register_external_die hook.
14035 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14036 (dwarf2_debug_hooks): Use them.
14037 (dwarf2_lineno_debug_hooks): Adjust.
14038 (struct die_struct): Add with_offset flag.
14039 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14040 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14041 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14042 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14043 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14044 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14045 defining section names for the early LTO debug variants.
14046 (reset_indirect_string): New helper.
14047 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14048 (print_dw_val): Add support for offsetted symbol references.
14049 (get_ultimate_context): Split out from is_cxx.
14050 (is_cxx): Use get_ultimate_context.
14051 (is_fortran): Add decl overload.
14052 (compute_comp_unit_symbol): Split out worker from
14053 compute_section_prefix.
14054 (compute_section_prefix): Call compute_comp_unit_symbol and
14055 set comdat_type_p here.
14056 (output_die): Skip DIE symbol output for the LTO added one.
14057 Handle DIE symbol references with offset.
14058 (output_comp_unit): Guard section name mangling properly.
14059 For LTO debug sections emit a symbol at the section beginning
14060 which we use to refer to its DIEs.
14061 (add_abstract_origin_attribute): For DIEs registered via
14062 dwarf2out_register_external_die directly refer to the early
14063 DIE rather than indirectly through the shadow one we created.
14064 Remove obsolete call to dwarf2out_abstract_function for
14065 non-function/block origins.
14066 (gen_array_type_die): When generating early LTO debug do
14067 not emit DW_AT_string_length.
14068 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14069 late when in LTO. As suggested place a gcc_unreachable for
14070 the DECL_ABSTRACT_P case.
14071 (gen_subprogram_die): Avoid another specification DIE
14072 for early built declarations/definitions for the late LTO case.
14073 (gen_variable_die): Add type references for late duplicated VLA dies
14075 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14076 we have the abstract instance already.
14077 (process_scope_var): Adjust decl DIE contexts in LTO which
14078 first puts them in limbo.
14079 (gen_decl_die): Do not generate type DIEs late apart from
14080 types for VLAs or for decls we do not yet have a DIE. Do not
14081 call dwarf2out_abstract_function late.
14082 (dwarf2out_early_global_decl): Make sure to create DIEs
14083 for abstract instances of a decl first.
14084 (dwarf2out_late_global_decl): Adjust comment.
14085 (output_macinfo_op): With multiple macro sections use
14086 macinfo_label_base to distinguish labels.
14087 (output_macinfo): Likewise. Update macinfo_label_base.
14088 Pass in the line info label.
14089 (note_variable_value_in_expr): When generating LTO resolve
14090 all variable values here by generating DIEs as needed.
14091 (init_sections_and_labels): Add early LTO debug flag parameter
14092 and generate different sections and names if set. Add generation
14093 counter for the labels so we can have multiple of them.
14094 (reset_dies): Helper to allow DIEs to be output multiple times.
14095 (dwarf2out_finish): When outputting DIEs to the fat part of an
14096 LTO object first reset DIEs.
14097 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14098 (modified_type_die): Check for decl_ultimate_origin being self
14100 (gen_type_die_with_usage): Likewise.
14101 (gen_typedef_die): Allow decl_ultimate_origin being self.
14102 (set_decl_abstract_flags): Remove.
14103 (set_block_abstract_flags): Likewise.
14104 (dwarf2out_abstract_function): Treat the early generated DIEs
14105 as the abstract copy and only add DW_AT_inline and
14106 DW_AT_artificial here and call set_decl_origin_self.
14107 If the DIE has an abstract origin don't do anything.
14108 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14109 if we have none yet (Go fails to build one, PR78628).
14110 (variably_modified_type_p): Prevent endless recursion for Ada
14111 cyclic pointer types.
14112 * lto-streamer-in.c: Include debug.h.
14113 (dref_queue): New global.
14114 (lto_read_tree_1): Stream in DIE references.
14115 (lto_input_tree): Register DIE references.
14116 (input_function): Stream DECL_DEBUG_ARGS.
14117 * lto-streamer-out.c: Include debug.h.
14118 (lto_write_tree_1): Output DIE references.
14119 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14120 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14121 (output_function): Stream DECL_DEBUG_ARGS.
14122 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14123 Stream DECL_ABSTRACT_ORIGIN.
14124 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14125 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14126 DECL_CONTEXT for file-scope decls.
14127 * lto-streamer.h (struct dref_entry): Declare.
14128 (dref_queue): Likewise.
14129 * cfgexpand.c (pass_expand::execute): Do not call the
14130 outlining_inline_function hook here.
14131 * lto-wrapper.c (debug_obj): New global.
14132 (tool_cleanup): Unlink it if required.
14133 (debug_objcopy): New function.
14134 (run_gcc): Handle early debug sections in the IL files by
14135 extracting them to separate files, partially linkin them and
14136 feeding the result back as result to the linker.
14137 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14138 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14139 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14140 sections into a separate segment.
14141 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14143 (darwin_asm_dwarf_section): Likewise.
14144 (darwin_asm_output_dwarf_offset): Likewise.
14145 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14147 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14148 Alan Hayward <alan.hayward@arm.com>
14149 David Sherwood <david.sherwood@arm.com>
14151 * read-md.h (md_reader::record_potential_iterator_use): Replace
14152 pointer argument with an rtx and an index.
14153 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14154 (apply_mode_iterator): Likewise.
14155 (apply_code_iterator): Likewise.
14156 (apply_int_iterator): Likewise.
14157 (apply_subst_iterator): Likewise.
14158 (record_iterator_use): Likewise.
14159 (record_attribute_use): Likewise.
14160 (md_reader::record_potential_iterator_use): Likewise. Update calls
14161 to record_iterator_use and apply_iterator.
14162 (iterator_use): Replace ptr with x and index.
14163 (attribute_use): Likewise.
14164 (apply_attribute_uses): Update calls to apply_iterator.
14165 (apply_iterators): Likewise. Update initialization of iterator_use.
14166 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14167 and record_potential_iterator_use.
14168 (rtx_reader::read_rtx_operand): Likewise.
14170 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14171 Alan Hayward <alan.hayward@arm.com>
14172 David Sherwood <david.sherwood@arm.com>
14174 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14177 2017-08-21 Richard Biener <rguenther@suse.de>
14179 PR middle-end/81884
14180 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14181 at struct end conservatively when comparing common bases.
14183 2017-08-21 Richard Biener <rguenther@suse.de>
14185 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14186 (mem_ref_in_stmt): Remove.
14187 (determine_max_movement): Use ref index to get at the reference.
14188 (invariantness_dom_walker::before_dom_children): Deal with
14189 lim data already initialized.
14190 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14192 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14194 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14195 (TARGET_ISA_ROUND): Ditto.
14196 (TARGET_ROUND): Ditto.
14197 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14198 * config/i386/i386.md: Ditto.
14199 * config/i386/sse.md: Ditto.
14200 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14201 with OPTION_MASK_ISA_SSE4_1.
14203 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14206 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14207 __builtin_ia32_lzcnt_u16.
14209 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14212 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14213 (rs6000_set_current_function): Rewrite function to use it.
14215 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14218 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14219 and TYPE_WARN_IF_NOT_ALIGN.
14220 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14221 (handle_warn_if_not_align): New.
14222 (place_union_field): Call handle_warn_if_not_align.
14223 (place_field): Call handle_warn_if_not_align.
14224 Copy TYPE_WARN_IF_NOT_ALIGN.
14225 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14226 (layout_type): Likewise.
14227 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14229 (tree_decl_common): Add warn_if_not_align.
14230 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14231 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14232 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14233 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14234 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14235 * doc/extend.texi: Document warn_if_not_aligned attribute.
14236 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14238 2017-08-17 Martin Liska <mliska@suse.cz>
14241 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14242 (get_data_dependence): Use it as pointer type.
14243 (distribute_loop): Likewise.
14245 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14247 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14248 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14249 (p8_vmrgow_<mode>_direct): New define_insn.
14250 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14251 handle endianness for vmrgew and vmrgow permute patterns.
14253 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14255 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14256 * config/rs6000/rs6000-cpus.def: Remove comment.
14257 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14258 (POWERPC_MASKS): Likewise.
14259 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14260 use of TARGET_VSX_TIMODE.
14261 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14262 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14263 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14264 (rs6000_option_override_internal): Remove dead code.
14265 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14266 (rs6000_legitimize_reload_address): Likewise.
14267 (rs6000_legitimate_address_p): Likewise.
14268 (rs6000_opt_masks): Delete "vsx-timode".
14269 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14270 from function comment.
14271 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14272 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14273 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14275 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14276 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14277 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14278 (VSX_TI): Likewise.
14280 (define_peephole2): Likewise.
14282 2017-08-17 Martin Sebor <msebor@redhat.com>
14285 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14286 past the end of an array.
14287 (test_pp_format): Add test cases.
14289 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14291 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14292 missing ECF_NOTHROW flags.
14294 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14297 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14298 operands residing in integer registers.
14299 (*vsx_le_perm_load_<mode>): Likewise.
14300 (*vsx_le_perm_store_<mode>): Likewise.
14301 (define_peephole2): Add peepholes to optimize the above.
14303 2017-08-17 Marek Polacek <polacek@redhat.com>
14305 PR middle-end/81814
14306 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14307 to mimic what shorten_compare did. Change the return type to bool.
14308 (fold_cond_expr_with_comparison): Update call to
14309 operand_equal_for_comparison_p.
14310 (fold_ternary_loc): Likewise.
14312 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14314 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14316 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14318 2017-08-17 Richard Biener <rguenther@suse.de>
14320 * tree-ssa-structalias.c (solve_graph): When propagating
14321 to successors update the graphs succ edges and avoid duplicate work.
14323 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14326 * config/i386/i386.c (ix86_option_override_internal): Save target
14327 specific options after ix86_stack_protector_guard_reg was changed.
14329 2017-08-17 Richard Biener <rguenther@suse.de>
14331 PR tree-optimization/81827
14332 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14333 (new_var_info): Initialize it conservatively.
14334 (get_call_vi): Mark register vars.
14335 (new_scalar_tmp_constraint_exp): Likewise.
14336 (handle_rhs_call): Likewise.
14337 (handle_const_call): Likewise.
14338 (create_function_info_for): Likewise.
14339 (solve_constraints): Sort varinfos to separate register from
14340 non-register vars to pack points-to solution bitmaps during
14343 2017-08-17 Marek Polacek <polacek@redhat.com>
14345 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14347 2017-08-17 Richard Biener <rguenther@suse.de>
14349 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14350 to true when overflow is undefined and we saturated the result.
14352 2017-08-17 Alan Modra <amodra@gmail.com>
14355 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14356 Don't use store multiple if only one reg needs saving.
14357 (interesting_frame_related_regno): New function.
14358 (rs6000_frame_related): Don't emit frame info for regs that
14360 (rs6000_emit_epilogue): Likewise.
14362 2017-08-16 Nathan Sidwell <nathan@acm.org>
14364 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14365 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14366 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14367 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14368 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14369 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14370 (verify_type): Adjust for TYPE_BINFO move.
14371 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14372 process TYPE_BINFO directly.
14373 (hash_tree): Likewise.
14374 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14376 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14379 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14381 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14383 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14386 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14387 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14388 Add ix86_binary_operator_ok to insn constraint.
14389 (*iordi_1_bts): Ditto.
14390 (*xordi_1_btc): Ditto.
14391 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14392 Update corresponding peephole2 pattern.
14396 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14398 PR tree-optimization/81832
14399 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14400 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14402 2017-08-16 Marek Polacek <polacek@redhat.com>
14405 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14406 regarding size_zero_node.
14408 2017-08-16 Martin Liska <mliska@suse.cz>
14411 * config.gcc: Respect previously set extra_objs in case
14414 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14416 PR tree-optimization/81835
14417 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14418 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14419 not depend on the phi.
14421 2017-08-16 Alan Modra <amodra@gmail.com>
14423 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14426 2017-08-16 Alan Modra <amodra@gmail.com>
14428 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14429 (save_reg_p): ..into this. Update all callers.
14430 (first_reg_to_save): Simplify.
14432 2017-08-16 Alan Modra <amodra@gmail.com>
14434 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14437 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14441 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14442 constant count if that count is less than 32.
14444 2017-08-15 Nathan Sidwell <nathan@acm.org>
14446 * gcc.c (execute): Emit friendlier message if inferior is killed
14447 by an external cause.
14449 2017-08-15 Richard Biener <rguenther@suse.de>
14451 PR tree-optimization/81790
14452 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14453 CONSTRUCTORs from simplifying and VN.
14455 2017-08-14 Martin Sebor <msebor@redhat.com>
14457 * builtin-attrs.def: Add comments.
14459 2017-08-14 Martin Sebor <msebor@redhat.com>
14462 * doc/extend.texi (attribute nonstring): Document new attribute.
14464 2017-08-14 Martin Sebor <msebor@redhat.com>
14467 * tree-diagnostic.c (default_tree_printer): Handle %G.
14468 * gimple-pretty-print.h (percent_G_format): Declare new function.
14469 * gimple-pretty-print.c (percent_G_format): Define.
14470 * tree-pretty-print.c (percent_K_format): Add argument.
14472 2017-08-14 Martin Sebor <msebor@redhat.com>
14474 PR translation/79998
14475 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14476 Remove a stray space.
14478 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14481 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14482 (*iordi_1_bts): Ditto.
14483 (*xordi_1_btc): Ditto.
14485 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14488 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14489 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14491 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14492 quoted strings, and make more translator-friendly.
14493 (darwin_rs6000_override_options): Likewise.
14494 (rs6000_option_override_internal): Likewise.
14495 (rs6000_return_in_memory): Fix overlong line.
14496 (init_cmulative_args): Use quoted strings, and make more
14497 translator-friendly.
14498 (rs6000_pass_by_reference): Fix overlong line.
14499 (def_builtin): Use quoted strings.
14500 (altivec_expand_predicate_builtin): Use quoted strings, and make
14501 more translator-friendly.
14502 (htm_expand_builtin): Use quoted strings.
14503 (cpu_expand_builtin): Use quoted strings, and make more
14504 translator-friendly.
14505 (altivec_expand_builtin): Likewise.
14506 (paired_expand_predicate_builtin): Likewise.
14507 (rs6000_invalid_builtin): Likewise.
14508 (builtin_function_type): Use quoted strings.
14509 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14510 more translator-friendly.
14511 (rs6000_trampoline_init): Likewise.
14512 (rs6000_handle_altivec_attribute): Likewise.
14513 (rs6000_inner_target_options): Use quoted strings.
14514 (rs6000_disable_incompatible_switches): Likewise.
14515 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14516 strings, and make more translator-friendly.
14517 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14519 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14521 PR tree-optimization/81799
14522 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14523 cond_expr to simple gimple operand.
14525 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14527 PR middle-end/46932
14528 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14530 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14534 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14535 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14536 TARGET_GASISR_PROLOGUES.
14537 * config/avr/avr.c (avr_option_override): Same.
14538 (avr_pass_pre_proep::execute): Same.
14540 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14543 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14544 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14546 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14548 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14549 define_insn_and_split. Split to a memory load from 0 in
14550 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14551 using PTR mode iterator.
14552 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14553 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14554 (*add_tp_<mode>): Redefine as define_insn_and_split.
14555 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14556 address space. Merge with *add_tp_x32 using PTR mode iterator.
14557 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14558 Split to an add with a memory load from 0 in
14559 DEFAULT_TLS_SEG_REG address space.
14561 2017-08-12 Andrew Pinski <apinski@cavium.com>
14563 * config/aarch64/aarch64-option-extensions.def (rdma):
14564 Fix feature string to what Linux prints out in /proc/cpuinfo.
14566 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14569 * dwarf2out.c (modified_type_die): For C typedef types that have
14570 an ultimate origin, process the ultimate origin instead of the
14572 (gen_typedef_die): Assert that input DECLs have no ultimate
14574 (gen_type_die_with_usage): For typedef variants that have an
14575 ultimate origin, just call gen_decl_die on the original DECL.
14576 (process_scope_var): Avoid creating DIEs for local typedefs and
14577 concrete static variables.
14579 2017-08-12 Alan Modra <amodra@gmail.com>
14583 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14584 match gnu-user.h startfile.
14585 (ENDFILE_LINUX_SPEC): Similarly.
14587 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14590 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14592 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14594 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14595 * config/aarch64/aarch64.md (mov<mode>): Change.
14596 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14597 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14598 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14600 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14602 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14603 for storage order barriers.
14605 2017-08-11 Martin Liska <mliska@suse.cz>
14608 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14609 variables of void type.
14611 2017-08-11 Martin Liska <mliska@suse.cz>
14613 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14614 TARGET_SUPPORTS_ALIASES.
14615 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14616 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14617 (optimize_weakref): Likewise.
14618 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14619 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14620 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14622 2017-08-11 Martin Liska <mliska@suse.cz>
14625 * config/i386/i386.c (make_resolver_func): Do complete
14626 refactoring of the function.
14628 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14631 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14632 * config/i386/i386.c (ix86_stack_protect_guard): Use
14633 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14634 * doc/invoke.texi (x86 Options): Document
14635 -mstack-protector-guard-symbol= option.
14637 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14639 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14640 * config/i386/i386.c (ix86_split_stack_guard): New function.
14641 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14642 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14643 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14644 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14645 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14646 (split_stack_space_check): Call ix86_split_stack_guard.
14648 2017-08-10 Martin Sebor <msebor@redhat.com>
14650 * print-tree.c (print_node): Print location using the established
14652 Replace spaces with colons.
14653 (debug_raw, debug): Ditto.
14655 2017-08-10 Martin Sebor <msebor@redhat.com>
14658 * pretty-print.c (pp_format): Correct the handling of %s precision.
14660 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14663 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14665 (ix86_finalize_stack_frame_flags): This. Also clear
14666 frame_pointer_needed if -fno-omit-frame-pointer is used without
14668 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14669 with ix86_finalize_stack_frame_flags.
14670 (ix86_expand_epilogue): Likewise.
14671 (ix86_expand_split_stack_prologue): Likewise.
14672 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14674 2017-08-10 Martin Liska <mliska@suse.cz>
14677 * c-attribs.c (handle_target_attribute):
14678 Report warning for an empty string argument of target attribute.
14680 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14683 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14685 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14686 or DECL_NONLOCAL labels.
14687 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14688 or DECL_NONLOCAL labels here.
14690 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14693 to indicate when early gimple folding has been disabled.
14694 (rs6000_gimple_fold_builtin): Add debug content.
14695 (rs6000_invalid_builtin): Fix whitespace.
14696 (rs6000_expand_builtin): Fix whitespace.
14697 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14699 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14702 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14703 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14705 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14707 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14708 (qdf24xx): Likewise.
14709 * config/aarch64/aarch64.md: Include falkor.md.
14710 * config/aarch64/falkor.md: New.
14712 2017-08-09 Marek Polacek <polacek@redhat.com>
14715 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14716 * diagnostic.c (emit_diagnostic): Add a comment.
14717 (emit_diagnostic_valist): New function.
14719 2017-08-09 Marek Polacek <polacek@redhat.com>
14722 * input.c (make_location): New overload.
14723 * input.h (make_location): Declare.
14725 2017-08-08 Alan Modra <amodra@gmail.com>
14726 H.J. Lu <hongjiu.lu@intel.com>
14729 * gcc.c (NO_PIE_SPEC): Delete.
14730 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14732 (LINK_PIE_SPEC): ..to here.
14733 (LINK_COMMAND_SPEC): Support -no-pie.
14734 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14735 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14737 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14738 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14739 (ENDFILE_CRTEND_SPEC): Similarly.
14741 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14744 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14745 (mstack-protector-guard-offset=): Ditto.
14746 * config/i386/i386.c (ix86_option_override): Handle
14747 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14749 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14750 ix86_stack_protect_guard_offset variables.
14751 (TARGET_STACK_PROTECT_GUARD): Always define.
14752 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14753 and -mstack-protector-guard-offset= options.
14755 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14757 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14758 boundary case for the last candidate.
14760 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14762 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14763 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14765 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14767 PR middle-end/19706
14768 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14769 * config/aarch64/aarch64-builtins.c
14770 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14771 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14772 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14774 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14775 Andrew Pinski <pinskia@gmail.com>
14777 PR middle-end/19706
14778 * internal-fn.def (XORSIGN): New.
14779 * optabs.def (xorsign_optab): New.
14780 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14781 (convert_expand_mult_copysign): New.
14782 (pass_optimize_widening_mul::execute): Call
14783 convert_expand_mult_copysign.
14785 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14787 PR tree-optimization/81354
14788 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14789 Insert on edges rather than explicitly creating landing pads.
14790 (analyze_candidates_and_replace): Commit edge inserts.
14792 2017-08-08 Richard Biener <rguenther@suse.de>
14794 PR middle-end/81719
14795 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14796 (expand_simple_operations): Also look through ADDR_EXPRs with
14797 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14799 2017-08-08 Richard Biener <rguenther@suse.de>
14801 PR tree-optimization/81723
14802 * tree-vect-slp.c (struct bst_traits): New hash traits.
14803 (bst_fail): New global.
14804 (vect_build_slp_tree_2): New worker, split out from ...
14805 (vect_build_slp_tree): ... this now wrapping it with using
14806 bst_fail set to cache SLP tree build fails. Properly handle
14808 (vect_analyze_slp_instance): Allocate and free bst_fail.
14810 2017-08-08 Martin Liska <mliska@suse.cz>
14813 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14814 LABEL_DECLs that can be from a different function.
14816 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14818 PR tree-optimization/81744
14819 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14820 loop's number of iterations.
14822 2017-08-08 Martin Liska <mliska@suse.cz>
14824 * asan.c: Include header files.
14825 * attribs.c (build_decl_attribute_variant): New function moved
14827 (build_type_attribute_qual_variant): Likewise.
14828 (cmp_attrib_identifiers): Likewise.
14829 (simple_cst_list_equal): Likewise.
14830 (omp_declare_simd_clauses_equal): Likewise.
14831 (attribute_value_equal): Likewise.
14832 (comp_type_attributes): Likewise.
14833 (build_type_attribute_variant): Likewise.
14834 (lookup_ident_attribute): Likewise.
14835 (remove_attribute): Likewise.
14836 (merge_attributes): Likewise.
14837 (merge_type_attributes): Likewise.
14838 (merge_decl_attributes): Likewise.
14839 (merge_dllimport_decl_attributes): Likewise.
14840 (handle_dll_attribute): Likewise.
14841 (attribute_list_equal): Likewise.
14842 (attribute_list_contained): Likewise.
14843 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14844 (lookup_attribute_by_prefix): Likewise.
14845 * bb-reorder.c: Include header files.
14846 * builtins.c: Likewise.
14847 * calls.c: Likewise.
14848 * cfgexpand.c: Likewise.
14849 * cgraph.c: Likewise.
14850 * cgraphunit.c: Likewise.
14851 * convert.c: Likewise.
14852 * dwarf2out.c: Likewise.
14853 * final.c: Likewise.
14854 * fold-const.c: Likewise.
14855 * function.c: Likewise.
14856 * gimple-expr.c: Likewise.
14857 * gimple-fold.c: Likewise.
14858 * gimple-pretty-print.c: Likewise.
14859 * gimple.c: Likewise.
14860 * gimplify.c: Likewise.
14861 * hsa-common.c: Likewise.
14862 * hsa-gen.c: Likewise.
14863 * internal-fn.c: Likewise.
14864 * ipa-chkp.c: Likewise.
14865 * ipa-cp.c: Likewise.
14866 * ipa-devirt.c: Likewise.
14867 * ipa-fnsummary.c: Likewise.
14868 * ipa-inline.c: Likewise.
14869 * ipa-visibility.c: Likewise.
14871 * lto-cgraph.c: Likewise.
14872 * omp-expand.c: Likewise.
14873 * omp-general.c: Likewise.
14874 * omp-low.c: Likewise.
14875 * omp-offload.c: Likewise.
14876 * omp-simd-clone.c: Likewise.
14877 * opts-global.c: Likewise.
14878 * passes.c: Likewise.
14879 * predict.c: Likewise.
14880 * sancov.c: Likewise.
14881 * sanopt.c: Likewise.
14882 * symtab.c: Likewise.
14883 * toplev.c: Likewise.
14884 * trans-mem.c: Likewise.
14885 * tree-chkp.c: Likewise.
14886 * tree-eh.c: Likewise.
14887 * tree-into-ssa.c: Likewise.
14888 * tree-object-size.c: Likewise.
14889 * tree-parloops.c: Likewise.
14890 * tree-profile.c: Likewise.
14891 * tree-ssa-ccp.c: Likewise.
14892 * tree-ssa-live.c: Likewise.
14893 * tree-ssa-loop.c: Likewise.
14894 * tree-ssa-sccvn.c: Likewise.
14895 * tree-ssa-structalias.c: Likewise.
14896 * tree-ssa.c: Likewise.
14897 * tree-streamer-in.c: Likewise.
14898 * tree-vectorizer.c: Likewise.
14899 * tree-vrp.c: Likewise.
14900 * tsan.c: Likewise.
14901 * ubsan.c: Likewise.
14902 * varasm.c: Likewise.
14903 * varpool.c: Likewise.
14904 * tree.c: Remove functions moved to attribs.[ch].
14905 * tree.h: Likewise.
14906 * config/aarch64/aarch64.c: Add attrs.h header file.
14907 * config/alpha/alpha.c: Likewise.
14908 * config/arc/arc.c: Likewise.
14909 * config/arm/arm.c: Likewise.
14910 * config/avr/avr.c: Likewise.
14911 * config/bfin/bfin.c: Likewise.
14912 * config/c6x/c6x.c: Likewise.
14913 * config/cr16/cr16.c: Likewise.
14914 * config/cris/cris.c: Likewise.
14915 * config/darwin.c: Likewise.
14916 * config/epiphany/epiphany.c: Likewise.
14917 * config/fr30/fr30.c: Likewise.
14918 * config/frv/frv.c: Likewise.
14919 * config/ft32/ft32.c: Likewise.
14920 * config/h8300/h8300.c: Likewise.
14921 * config/i386/winnt.c: Likewise.
14922 * config/ia64/ia64.c: Likewise.
14923 * config/iq2000/iq2000.c: Likewise.
14924 * config/lm32/lm32.c: Likewise.
14925 * config/m32c/m32c.c: Likewise.
14926 * config/m32r/m32r.c: Likewise.
14927 * config/m68k/m68k.c: Likewise.
14928 * config/mcore/mcore.c: Likewise.
14929 * config/microblaze/microblaze.c: Likewise.
14930 * config/mips/mips.c: Likewise.
14931 * config/mmix/mmix.c: Likewise.
14932 * config/mn10300/mn10300.c: Likewise.
14933 * config/moxie/moxie.c: Likewise.
14934 * config/msp430/msp430.c: Likewise.
14935 * config/nds32/nds32-isr.c: Likewise.
14936 * config/nds32/nds32.c: Likewise.
14937 * config/nios2/nios2.c: Likewise.
14938 * config/nvptx/nvptx.c: Likewise.
14939 * config/pa/pa.c: Likewise.
14940 * config/pdp11/pdp11.c: Likewise.
14941 * config/powerpcspe/powerpcspe.c: Likewise.
14942 * config/riscv/riscv.c: Likewise.
14943 * config/rl78/rl78.c: Likewise.
14944 * config/rx/rx.c: Likewise.
14945 * config/s390/s390.c: Likewise.
14946 * config/sh/sh.c: Likewise.
14947 * config/sol2.c: Likewise.
14948 * config/sparc/sparc.c: Likewise.
14949 * config/spu/spu.c: Likewise.
14950 * config/stormy16/stormy16.c: Likewise.
14951 * config/tilegx/tilegx.c: Likewise.
14952 * config/tilepro/tilepro.c: Likewise.
14953 * config/v850/v850.c: Likewise.
14954 * config/vax/vax.c: Likewise.
14955 * config/visium/visium.c: Likewise.
14956 * config/xtensa/xtensa.c: Likewise.
14958 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14961 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14962 constraints since the -mupper-regs-* switches have been
14964 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14965 into a vector from a double word element that was extracted from
14966 another vector, and eliminate extra XXPERMDI instructions.
14967 (vsx_concat_<mode>_2): Likewise.
14968 (vsx_concat_<mode>_3): Likewise.
14969 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14970 concat to allow optimizing inserts from previous extracts.
14972 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14974 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14975 memory reference to a SSP offset in TLS address space.
14976 (ix86_print_operand) <case '@'>: Remove.
14977 (ix86_print_operand_punct_valid_p): Remove '@' code.
14978 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14979 UNSPEC_SP_TLS_TEST.
14980 (stack_tls_protect_set_<mode>): Remove.
14981 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14982 (stack_tls_protect_test_<mode>): Remove.
14983 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14985 2017-08-07 Olivier Hainque <hainque@adacore.com>
14988 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14990 2017-08-07 Douglas Rupp <rupp@adacore.com>
14992 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14993 variable was referenced as multidir in command.
14995 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14998 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15001 2017-08-07 Martin Liska <mliska@suse.cz>
15003 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15004 * config/rl78/rl78.c: Add include of attribs.h.
15005 * config/sh/sh.c: Likewise.
15006 * config/v850/v850.c: Likewise.
15008 2017-08-07 Tom de Vries <tom@codesourcery.com>
15010 PR middle-end/78266
15011 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15013 2017-08-07 Martin Liska <mliska@suse.cz>
15015 * config/mips/mips.c: Include attribs.h.
15017 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15020 * doc/invoke.texi: Document change in behvaior for -Ofast for
15023 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15025 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15027 (aarch64_pop_regs): Likewise.
15028 (aarch64_gen_load_pair): Likewise.
15029 (aarch64_save_callee_saves): Likewise.
15030 (aarch64_restore_callee_saves): Likewise.
15032 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15034 * config/i386/i386.c: Revert the last change.
15036 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15039 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15041 (ix86_finalize_stack_frame_flags): This. Also clear
15042 frame_pointer_needed if -fno-omit-frame-pointer is used without
15044 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15045 with ix86_finalize_stack_frame_flags.
15046 (ix86_expand_epilogue): Likewise.
15047 (ix86_expand_split_stack_prologue): Likewise.
15049 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15052 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15053 to P_AES for Westmere.
15055 2017-08-07 Jonathan Yong <10walls@gmail.com>
15057 * config/i386/mingw.opt (fset-stack-executable): Removed.
15058 * config/i386/cygming.opt (fset-stack-executable): Moved
15060 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15062 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15064 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15066 2017-08-07 Marek Polacek <polacek@redhat.com>
15068 PR middle-end/81737
15069 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15071 2017-08-07 Martin Liska <mliska@suse.cz>
15073 * attribs.h (canonicalize_attr_name): New function.
15074 (cmp_attribs): Move from c-format.c and adjusted.
15075 (is_attribute_p): Moved from tree.h.
15076 * tree-inline.c: Add new includes.
15077 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15078 (private_is_attribute_p): Remove.
15079 (private_lookup_attribute): Likewise.
15080 (private_lookup_attribute_by_prefix): Simplify.
15081 (remove_attribute): Use is_attribute_p.
15082 * tree.h: Remove removed declarations.
15084 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15086 PR middle-end/81698
15087 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15088 instead of computing it in the function. Formatting fix.
15089 (expand_case): Don't rely on default_edge being the first edge,
15090 clear it if removing it, pass default_edge to
15091 emit_case_dispatch_table.
15092 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15095 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15097 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15098 insn in the function, emit NOP after the insn.
15100 2017-08-06 Tom de Vries <tom@codesourcery.com>
15102 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15105 2017-08-06 Tom de Vries <tom@codesourcery.com>
15107 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15110 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15112 PR tree-optimization/57371
15113 * match.pd: New pattern.
15115 2017-08-04 Marek Polacek <polacek@redhat.com>
15117 PR middle-end/81695
15118 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15119 perform the computation in offset_int.
15121 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15123 PR tree-optimization/81136
15124 * tree-vectorizer.h: Include tree-hash-traits.h.
15125 (vec_base_alignments): New typedef.
15126 (vec_info): Add a base_alignments field.
15127 (vect_record_base_alignments): Declare.
15128 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15130 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15131 (create_data_ref): Add an is_conditional_in_stmt argument.
15132 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15133 the is_conditional_in_stmt field.
15134 (data_ref_loc): Add an is_conditional_in_stmt field.
15135 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15136 (find_data_references_in_stmt): Update call to create_data_ref.
15137 (graphite_find_data_references_in_stmt): Likewise.
15138 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15139 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15140 (vect_record_base_alignment): New function.
15141 (vect_record_base_alignments): Likewise.
15142 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15143 for nested statements even if we fail to compute a misalignment.
15144 Use pooled base alignments for unconditional references.
15145 (vect_find_same_alignment_drs): Compare base addresses instead
15147 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15148 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15150 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15152 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15153 Add an explicit name for the enum. Use auto_vec for slp_instances
15154 and grouped_stores.
15155 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15157 (_bb_vec_info): Add a constructor and destructor.
15158 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15159 (destroy_loop_vec_info): Delete.
15160 (vect_destroy_datarefs): Likewise.
15161 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15162 (vec_info::vec_info): New function.
15163 (vec_info::~vec_info): Likewise.
15164 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15165 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15166 destroy_loop_vec_info.
15167 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15168 (_loop_vec_info::_loop_vec_info): ...this.
15169 (destroy_loop_vec_info): Replace with...
15170 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15171 the stmt_vec_infos. Leave handling of vec_info information to its
15172 destructor. Remove explicit vector releases.
15173 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15174 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15175 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15176 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15177 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15178 (destroy_bb_vec_info): Replace with...
15179 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15180 information to its destructor.
15181 (vect_slp_analyze_bb_1): Use new and delete instead of
15182 new_bb_vec_info and destroy_bb_vec_info.
15183 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15186 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15188 * tree-data-ref.h (subscript): Add access_fn field.
15189 (data_dependence_relation): Add could_be_independent_p.
15190 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15191 (same_access_functions): Move to tree-data-ref.c.
15192 * tree-data-ref.c (ref_contains_union_access_p): New function.
15193 (access_fn_component_p): Likewise.
15194 (access_fn_components_comparable_p): Likewise.
15195 (dr_analyze_indices): Add a reference to access_fn_component_p.
15196 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15198 (constant_access_functions): Likewise.
15199 (add_other_self_distances): Likewise.
15200 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15201 (initialize_data_dependence_relation): Use XCNEW and remove
15202 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15203 of access functions that have the same type. Allow the
15204 subsequence to end with different bases in some circumstances.
15205 Record the chosen access functions in SUB_ACCESS_FN.
15206 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15207 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15208 (subscript_dependence_tester_1): Likewise dra and drb.
15209 (build_classic_dist_vector): Update calls accordingly.
15210 (subscript_dependence_tester): Likewise.
15211 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15212 DDR_COULD_BE_INDEPENDENT_P.
15213 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15214 comp_alias_ddrs instead of may_alias_ddrs.
15215 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15217 (vect_analyze_data_ref_dependence): Use it if
15218 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15219 distance vectors if that fails.
15220 (dependence_distance_ge_vf): New function.
15221 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15222 LOOP_VINFO_MAY_ALIAS_DDRS.
15224 2017-08-04 Richard Biener <rguenther@suse.de>
15226 PR middle-end/81705
15227 * fold-const.c (fold_binary_loc): Properly restrict
15228 minus_var0 && minus_var1 case when associating undefined overflow
15231 2017-08-04 Tom de Vries <tom@codesourcery.com>
15233 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15235 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15238 Don't start diagnostic messages with a capital letter.
15239 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15241 (rs6000_invalid_builtin): Likewise.
15242 (rs6000_trampoline_init): Likewise.
15244 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15247 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15248 after setting changeable df flags.
15250 2017-08-03 Richard Biener <rguenther@suse.de>
15252 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15253 up if the use is in USE - X.
15255 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15257 * toplev.c (dumpfile.h): New include.
15258 (internal_error_reentered): New static function. Use it...
15259 (internal_error_function): ...here to handle reentered internal_error.
15261 2017-08-03 Richard Biener <rguenther@suse.de>
15263 PR middle-end/81148
15264 * fold-const.c (split_tree): Add minus_var and minus_con
15265 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15266 here but always use minus_*.
15267 (associate_trees): Assert we never associate with MINUS_EXPR
15268 and NULL first operand. Do not recurse for PLUS_EXPR operands
15269 when associating as MINUS_EXPR either.
15270 (fold_binary_loc): Track minus_var and minus_con.
15272 2017-08-03 Tom de Vries <tom@codesourcery.com>
15275 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15276 ACCEL_COMPILER, apply finish_options on
15277 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15279 2017-08-03 Tom de Vries <tom@codesourcery.com>
15282 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15283 function_entry_patch_area_size > 0.
15285 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15288 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15289 instead of 10??LU, perform unit multiplication in wide_int,
15290 don't change alloc_object_size_limit if the limit is larger
15293 PR tree-optimization/81655
15294 PR tree-optimization/81588
15295 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15296 the case when ranges[i].low and high are 1 for unsigned type with
15299 PR middle-end/81052
15300 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15301 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15303 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15305 * tree-vrp.h: Add include guard.
15307 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15310 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15311 (ud2): New insn pattern.
15312 * config/i386/i386.c (ix86_expand_epilogue):
15313 For naked functions, generate ud2 instead of trap insn.
15315 2017-08-02 Marek Polacek <polacek@redhat.com>
15318 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15320 2017-08-02 Tom de Vries <tom@codesourcery.com>
15321 Cesar Philippidis <cesar@codesourcery.com>
15323 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15324 Add missing edge probabilities.
15326 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15328 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15329 Correct endianness.
15331 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15333 PR middle-end/79499
15334 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15335 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15336 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15337 prologue_seq sequences - if any.
15339 2017-08-02 Richard Biener <rguenther@suse.de>
15341 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15342 via vectors if supported, integer extracts via punning if supported
15343 or otherwise vector extracts.
15345 2017-08-02 Richard Biener <rguenther@suse.de>
15347 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15349 (bitmap_insert_into_set): ... this.
15351 2017-08-02 Richard Biener <rguenther@suse.de>
15353 PR tree-optimization/81633
15355 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15357 PR tree-optimization/71752
15358 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15360 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15362 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15363 (machine_function::call_ms2sysv_pad_out): Remove field.
15364 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15365 (ix86_compute_frame_layout): Likewise.
15367 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15370 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15371 attribute with interrupt attribute.
15373 2017-08-01 Andrew Pinski <apinski@cavium.com>
15375 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15376 BIT_INSERT_EXPR's operand 1
15377 to see if the types precision matches.
15379 2017-08-01 Martin Liska <mliska@suse.cz>
15381 PR middle-end/70140
15382 * builtins.c (expand_builtin_memcpy_args): Remove.
15383 (expand_builtin_memcpy): Call newly added function
15384 expand_builtin_memory_copy_args.
15385 (expand_builtin_memcpy_with_bounds): Likewise.
15386 (expand_builtin_mempcpy): Remove last argument.
15387 (expand_builtin_mempcpy_with_bounds): Likewise.
15388 (expand_builtin_memory_copy_args): New function created from
15389 expand_builtin_mempcpy_args with small modifications.
15390 (expand_builtin_mempcpy_args): Remove.
15391 (expand_builtin_stpcpy): Remove unused argument.
15392 (expand_builtin): Likewise.
15393 (expand_builtin_with_bounds): Likewise.
15395 2017-08-01 Martin Liska <mliska@suse.cz>
15398 Make mempcpy more optimal (PR middle-end/70140).
15400 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15403 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15404 __builtin_vec_cmpne verify both arguments are compatible vectors
15405 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15406 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15407 move computation of aligned to after checking the argument types.
15411 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15412 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15415 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15416 Doug Rupp <rupp@adacore.com>
15417 Olivier Hainque <hainque@adacore.com>
15419 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15420 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15422 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15423 for TARGET_OS_CPP_BUILTIN.
15424 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15425 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15427 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15428 passing required abi options to the assembler for EABI configurations.
15429 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15430 of .text.hot and .text.unlikely sections for kernel modules when
15431 using ARM style exceptions.
15432 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15433 options. Add EXTRA_CC1_SPEC.
15434 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15436 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15438 (ARM_TARGET2_DWARF_FORMAT): Define.
15439 * config/arm/t-vxworks: Adjust multilib control to removal of the
15440 Diab command line options.
15442 2017-08-01 Martin Liska <mliska@suse.cz>
15444 PR gcov-profile/81561
15445 * gcov.c (unblock): Make unblocking safe as we need to preserve
15446 index correspondence of blocks and block_lists.
15448 2017-08-01 Richard Biener <rguenther@suse.de>
15450 PR tree-optimization/81181
15451 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15452 (compute_antic): ... end of iteration here.
15454 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15456 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15457 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15458 (ftree-slp-vectorize): Likewise.
15459 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15460 can no longer be set independent of flag_tree_loop_vectorize.
15461 * omp-general.c (emp_max_vf): Likewise.
15462 * opts.c (enable_fdo_optimizations): Remove references to
15463 flag_tree_vectorize, these are now implicit.
15464 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15465 and leave it for the options machinery.
15467 2017-08-01 Martin Liska <mliska@suse.cz>
15469 PR middle-end/70140
15470 * builtins.c (expand_builtin_memcpy_args): Remove.
15471 (expand_builtin_memcpy): Call newly added function
15472 expand_builtin_memory_copy_args.
15473 (expand_builtin_memcpy_with_bounds): Likewise.
15474 (expand_builtin_mempcpy): Remove last argument.
15475 (expand_builtin_mempcpy_with_bounds): Likewise.
15476 (expand_builtin_memory_copy_args): New function created from
15477 expand_builtin_mempcpy_args with small modifications.
15478 (expand_builtin_mempcpy_args): Remove.
15479 (expand_builtin_stpcpy): Remove unused argument.
15480 (expand_builtin): Likewise.
15481 (expand_builtin_with_bounds): Likewise.
15483 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15486 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15487 print "ds:" only for immediates in generic address space.
15489 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15492 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15493 (ix86_function_ok_for_sibcall): Return false for naked functions.
15495 2017-08-01 Richard Biener <rguenther@suse.de>
15497 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15498 (compute_antic): Seed worklist with exit block predecessors.
15499 * cfganal.c (dfs_find_deadend): For a cycle return the source
15500 of the edge closing it.
15502 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15504 * config/aarch64/aarch64.c
15505 (aarch64_can_const_movi_rtx_p): Move 0 check.
15507 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15509 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15510 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15512 * match.pd: Ditto in address comparison pattern.
15514 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15516 PR tree-optimization/81627
15517 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15518 closed ssa form for store-store chain.
15520 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15522 PR tree-optimization/81620
15523 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15524 for store-store chain.
15526 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15528 PR tree-optimization/81588
15529 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15530 ranges[i].in_p, invert comparison code ccode. For >/>=,
15531 swap rhs1 and rhs2 and comparison code unconditionally,
15532 for </<= don't do that. Don't swap rhs1/rhs2 again if
15533 ranges[i].in_p, instead invert comparison code ccode if
15534 opcode or oe->rank is BIT_IOR_EXPR.
15537 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15538 a direct optab to conversion optab.
15539 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15540 with GET_MODE_INNER as last argument instead of optab_handler.
15541 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15542 vector extraction if possible and optab is available.
15543 * expr.c (store_constructor): Use convert_optab_handler instead
15544 of optab_handler. Use vector initialization from smaller
15545 vectors if possible and optab is available.
15546 * tree-vect-stmts.c (vectorizable_load): Likewise.
15547 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15548 now have two modes.
15549 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15550 of vec_init from half-sized vectors with the same element mode.
15551 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15552 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15553 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15554 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15555 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15556 after mode in gen_vec_extract* calls.
15557 (vec_extract<mode>): Renamed to ...
15558 (vec_extract<mode><ssescalarmodelower>): ... this.
15559 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15560 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15561 element mode after mode in gen_vec_init* calls.
15562 (VEC_INIT_HALF_MODE): New mode iterator.
15563 (vec_init<mode>): Renamed to ...
15564 (vec_init<mode><ssescalarmodelower>): ... this.
15565 (vec_init<mode><ssehalfvecmodelower>): New expander.
15566 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15567 (vec_extractv2sfsf): ... this.
15568 (vec_initv2sf): Renamed to ...
15569 (vec_initv2sfsf): ... this.
15570 (vec_extractv2si): Renamed to ...
15571 (vec_extractv2sisi): ... this.
15572 (vec_initv2si): Renamed to ...
15573 (vec_initv2sisi): ... this.
15574 (vec_extractv4hi): Renamed to ...
15575 (vec_extractv4hihi): ... this.
15576 (vec_initv4hi): Renamed to ...
15577 (vec_initv4hihi): ... this.
15578 (vec_extractv8qi): Renamed to ...
15579 (vec_extractv8qiqi): ... this.
15580 (vec_initv8qi): Renamed to ...
15581 (vec_initv8qiqi): ... this.
15582 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15583 (vec_init<mode>): Renamed to ...
15584 (vec_init<mode><VEC_base_l>): ... this.
15585 (vec_extract<mode>): Renamed to ...
15586 (vec_extract<mode><VEC_base_l>): ... this.
15587 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15588 (vec_initv2sfsf): ... this.
15589 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15590 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15591 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15592 element mode after mode in gen_vec_init* calls.
15593 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15594 (vec_init<mode><Vel>): ... this.
15595 (vec_extract<mode>): Renamed to ...
15596 (vec_extract<mode><Vel>): ... this.
15597 * config/aarch64/iterators.md (Vel): New mode attribute.
15598 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15599 Add element mode after mode in gen_vec_extract* calls.
15600 * config/s390/vector.md (non_vec_l): New mode attribute.
15601 (vec_extract<mode>): Renamed to ...
15602 (vec_extract<mode><non_vec_l>): ... this.
15603 (vec_init<mode>): Renamed to ...
15604 (vec_init<mode><non_vec_l>): ... this.
15605 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15606 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15608 * config/arm/iterators.md (V_elem_l): New mode attribute.
15609 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15610 (vec_extract<mode><V_elem_l>): ... this.
15611 (vec_extractv2di): Renamed to ...
15612 (vec_extractv2didi): ... this.
15613 (vec_init<mode>): Renamed to ...
15614 (vec_init<mode><V_elem_l>): ... this.
15615 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15616 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15617 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15618 Add element mode after gen_vec_extract* calls.
15619 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15620 (vec_init<mode><unitmode>): ... this.
15621 (vec_extract<mode>): Renamed to ...
15622 (vec_extract<mode><unitmode>): ... this.
15623 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15624 (vec_init<mode><unitmode>): ... this.
15625 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15626 (vec_initv2sfsf): ... this.
15627 (vec_extractv2sf): Renamed to ...
15628 (vec_extractv2sfsf): ... this.
15629 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15630 Add element mode after gen_vec_extract* calls.
15631 * config/mips/mips.md (unitmode): New mode iterator.
15632 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15633 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15634 * config/spu/spu.md (inner_l): New mode attribute.
15635 (vec_init<mode>): Renamed to ...
15636 (vec_init<mode><inner_l>): ... this.
15637 (vec_extract<mode>): Renamed to ...
15638 (vec_extract<mode><inner_l>): ... this.
15639 * config/sparc/sparc.md (veltmode): New mode iterator.
15640 (vec_init<VMALL:mode>): Renamed to ...
15641 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15642 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15643 (vec_initv2sisi): ... this.
15644 (vec_initv2sf): Renamed to ...
15645 (vec_initv2sfsf): ... this.
15646 (vec_extractv2sf): Renamed to ...
15647 (vec_extractv2sfsf): ... this.
15648 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15649 (vec_init<mode>): Renamed to ...
15650 (vec_init<mode><VEC_base_l>): ... this.
15651 (vec_extract<mode>): Renamed to ...
15652 (vec_extract<mode><VEC_base_l>): ... this.
15653 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15654 (vec_initv2sfsf): ... this.
15655 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15656 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15657 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15658 gen_vec_init* calls.
15660 2017-08-01 Richard Biener <rguenther@suse.de>
15662 PR tree-optimization/81297
15663 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15664 TREE_OVERFLOW from INTEGER_CSTs.
15666 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15668 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15670 2017-07-31 Carl Love <cel@us.ibm.com>
15672 * config/rs6000/rs6000-c: Add support for built-in functions
15673 vector signed char vec_xl_be (signed long long, signed char *);
15674 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15675 vector signed int vec_xl_be (signed long long, signed int *);
15676 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15677 vector signed long long vec_xl_be (signed long long, signed long long *);
15678 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15679 vector signed short vec_xl_be (signed long long, signed short *);
15680 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15681 vector double vec_xl_be (signed long long, double *);
15682 vector float vec_xl_be (signed long long, float *);
15683 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15684 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15685 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15687 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15688 (altivec_expand_builtin): Add switch statement to call
15689 altivec_expand_xl_be for each builtin.
15690 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15691 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15692 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15693 __builtin_vsx_le_be_v16qi.
15694 * doc/extend.texi: Update the built-in documentation file for the
15695 new built-in functions.
15697 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15700 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15702 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15704 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15706 * config.gcc: Add z14.
15707 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15708 CPU model numbers for z13s and z14.
15709 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15711 * config/s390/s390-opts.h (enum processor_type): Rename
15712 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15713 * config/s390/s390.c (processor_table): Add field for CPU name to
15714 be passed to Binutils.
15715 (s390_asm_output_machine_for_arch): Use the new field in
15716 processor_table for Binutils.
15717 (s390_expand_builtin): Replace arch12 with z14.
15718 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15719 (s390_get_sched_attrmask): Likewise.
15720 (s390_get_unit_mask): Likewise.
15721 * config/s390/s390.opt: Add z14 to processor_type enum.
15723 2017-07-31 Martin Jambor <mjambor@suse.cz>
15726 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15727 regardless of optimization level.
15729 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15730 Martin Liska <mliska@suse.cz>
15732 * predict.def: Remove old comment and adjust probability.
15733 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15734 PREDICT statements.
15736 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15739 * config/i386/i386.c (ix86_function_naked): New function.
15740 (ix86_can_use_return_insn_p): Return false for naked functions.
15741 (ix86_expand_prologue): Skip prologue for naked functions.
15742 (ix86_expand_epilogue): Skip epilogue for naked functions
15743 and emit trap instruction.
15744 (ix86_warn_func_return): New function.
15745 (ix86_attribute_table): Add "naked" attribute specification.
15746 (TARGET_WARN_FUNC_RETURN): Define.
15747 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15749 2017-07-31 Martin Liska <mliska@suse.cz>
15751 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15752 (dump_gimple_bb_header): Always dump BB info.
15753 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15755 2017-07-31 Martin Liska <mliska@suse.cz>
15758 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15759 also with current_function_decl non-null equality.
15761 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15764 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15765 change type to the element type, instead add eltype variable and
15766 use it where we are interested in the element type.
15768 PR tree-optimization/81603
15769 * ipa-polymorphic-call.c
15770 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15771 offset arithmetic in offset_int, bail out if the resulting bit offset
15772 doesn't fit into shwi.
15774 2017-07-31 Martin Liska <mliska@suse.cz>
15776 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15777 (gimplify_save_expr): Fix comment.
15779 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15782 * config/i386/i386.c (ix86_function_arg): Update arguments for
15784 (ix86_compute_frame_layout): Set the initial stack offset to
15785 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15786 INCOMING_FRAME_SP_OFFSET.
15787 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15788 stack before exception handler returns.
15789 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15790 the 'ERROR_CODE' for exception handler.
15792 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15794 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15795 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15796 (ASM_OUTPUT_REG_POP): Ditto.
15797 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15798 instead of asm_fprintf to output pure string.
15800 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15802 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15803 to imported_module_or_decl hook.
15804 (debug_nothing_tree_tree_tree_bool): Remove.
15805 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15806 * debug.c (do_nothing_debug_hooks): Use
15807 debug_nothing_tree_tree_tree_bool_bool instead of
15808 debug_nothing_tree_tree_tree_bool.
15809 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15810 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15811 * sdbout.c (sdb_debug_hooks): Likewise.
15812 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15813 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15815 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15816 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15817 attribute, don't add anything.
15819 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15821 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15822 (fold_build2_stat_loc): Likewise.
15823 (fold_build3_stat_loc): Likewise.
15824 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15825 (fold_build1_loc): Remove macro.
15826 (fold_build2_loc): Likewise.
15827 (fold_build3_loc): Likewise.
15829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15831 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15832 (gimple_build_debug_bind_source_stat): Likewise.
15833 * gimple.h (gimple_build_debug_bind): Remove macro.
15834 (gimple_build_debug_bind_source): Likewise.
15836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15838 * bitmap.c (bitmap_alloc): Adjust.
15839 (bitmap_gc_alloc): Likewise.
15840 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15844 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15845 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15846 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15847 (bitmap_gc_alloc_stat): Likewise.
15848 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15852 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15853 * rtl.h (shallow_copy_rtx): Remove macro.
15855 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15857 * emit-rtl.c (gen_raw_REG): Adjust.
15858 * gengenrtl.c (gendef): Likewise.
15859 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15860 * rtl.h (rtx_alloc): Remove macro.
15862 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15864 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15865 (build_tree_list_stat): Likewise.
15866 * tree.h (build_tree_list): Remove macro.
15867 (build_tree_list_vec): Likewise.
15869 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15871 * tree.c (make_vector_stat): Remove _stat from name.
15872 (build_vector_stat): Likewise.
15873 * tree.h (make_vector_stat): Remove macro.
15874 (build_vector_stat): Likewise.
15876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15878 * tree.h (build_var_debug_value): Remove prototype.
15880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15882 * tree.c (tree_cons_stat): Remove _stat from name.
15883 * tree.h (tree_cons): Remove macro.
15885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15887 * tree.c (build_vl_exp_stat): Remove _stat from name.
15888 * tree.h (build_vl_exp): Remove macro.
15890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15892 * tree.c (build_decl_stat): Remove _stat from name.
15893 * tree.h (build_decl): Remove macro.
15895 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15897 * gimple.c (gimple_build_with_ops_stat): Adjust.
15898 (gimple_alloc_stat): Remove _stat from name.
15899 * gimple.h (gimple_alloc): Remove macro.
15901 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15903 * tree.c (make_tree_vec_stat): Remove _stat from name.
15904 (grow_tree_vec_stat): Likewise.
15905 * tree.h (make_tree_vec_stat): Adjust prototype.
15906 (grow_tree_vec_stat): Likewise.
15907 (make_tree_vec): Remove macro.
15908 (grow_tree_vec): Likewise.
15910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15912 * fold-const.c (fold_build1_stat_loc): Adjust.
15913 (fold_build2_stat_loc): Likewise.
15914 (fold_build3_stat_loc): Likewise.
15915 * tree.c (build0_stat): Remove _stat from name.
15916 (build1_stat): Likewise.
15917 (build2_stat): Likewise.
15918 (build3_stat): Likewise.
15919 (build4_stat): Likewise.
15920 (build5_stat): Likewise.
15921 * tree.h (build1_loc): Remove macro, and rename _stat function
15923 (build2_loc): Likewise.
15924 (build3_loc): Likewise.
15925 (build4_loc): Likewise.
15926 (build5_loc): Likewise.
15928 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15930 * tree.c (make_int_cst_stat): Remove _stat from name.
15931 * tree.h (make_int_cst_stat): Adjust prototype.
15932 (make_int_cst): Remove macro.
15934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15936 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15937 * tree.h (make_tree_binfo_stat): Adjust prototype.
15938 (make_tree_binfo): Remove.
15940 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15942 * tree.c (copy_node_stat): Rename to copy_node.
15943 (build_distinct_type_copy): Adjust.
15944 * tree.h (copy_node_stat): Adjust prototype.
15945 (copy_node): Remove macro.
15947 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15949 * tree.c (make_node_stat): rename to make_node.
15950 (build_tree_list_stat): Adjust.
15951 (build0_stat): Likewise.
15952 (build2_stat): Likewise.
15953 (build3_stat): Likewise.
15954 (build4_stat): Likewise.
15955 (build5_stat): Likewise.
15956 (build_decl_stat): Likewise.
15957 * tree.h (make_node_stat): Adjust prototype.
15958 (make_node): remove macro.
15960 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15962 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15963 (PPC_FEATURE2_SCV): Likewise.
15964 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15966 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15968 * config/aarch64/aarch64.c
15969 (aarch64_internal_mov_immediate): Add new special pattern.
15970 * config/aarch64/aarch64.md (*movdi_aarch64):
15971 Add reg/32bit const mov case.
15973 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15974 Richard Sandiford <richard.sandiford@linaro.org>
15976 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15977 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15978 Add integer and movi cases.
15979 (movi-split-hf-df-sf split, fp16): New.
15980 (enabled): Added TARGET_FP_F16INST.
15981 * config/aarch64/iterators.md (GPF_HF): New.
15982 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15984 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15986 * config/aarch64/aarch64.c
15987 (aarch64_simd_container_mode): Add prototype.
15988 (aarch64_expand_mov_immediate): Add HI support.
15989 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15990 (aarch64_can_const_movi_rtx_p): New.
15991 (aarch64_preferred_reload_class):
15992 Remove restrictions of using FP registers for certain SIMD operations.
15993 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15994 (aarch64_valid_floating_const): Add integer move validation.
15995 (aarch64_simd_imm_scalar_p): Remove.
15996 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15997 (aarch64_legitimate_constant_p): Expand list of supported cases.
15998 * config/aarch64/aarch64-protos.h
15999 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16000 (aarch64_reinterpret_float_as_int): New.
16001 (aarch64_simd_imm_scalar_p): Remove.
16002 * config/aarch64/constraints.md (Uvi): New.
16003 (Dd): Split into Ds and new Dd.
16004 * config/aarch64/aarch64.md (*movsi_aarch64):
16006 (*movdi_aarch64): Add SIMD mov case.
16008 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16010 * tree-predcom.c: (struct chain): Handle store-store chain in which
16011 stores for elimination only store loop invariant values.
16012 (execute_pred_commoning_chain): Ditto.
16013 (prepare_initializers_chain_store_elim): Ditto.
16014 (prepare_finalizers): Ditto.
16015 (is_inv_store_elimination_chain): New function.
16016 (initialize_root_vars_store_elim_1): New function.
16018 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16020 * tree-predcom.c: Revise general description of the pass.
16021 (enum chain_type): New enum type for store elimination.
16022 (struct chain): New field supporting store elimination.
16023 (struct component): Ditto.
16024 (dump_chain): Dump store-stores chain.
16025 (release_chain): Release resources.
16026 (split_data_refs_to_components): Compute and create component
16027 contains only stores for elimination.
16028 (get_chain_last_ref_at): New function.
16029 (make_invariant_chain): Initialization.
16030 (make_rooted_chain): Specify chain type in parameter and record it.
16031 (add_looparound_copies): Skip for store-stores chain.
16032 (determine_roots_comp): Compute type of chain and pass it to
16034 (initialize_root_vars_store_elim_2): New function.
16035 (finalize_eliminated_stores): New function.
16036 (remove_stmt): Handle store for elimination.
16037 (execute_pred_commoning_chain): Execute predictive commoning on
16038 store-store chains.
16039 (determine_unroll_factor): Skip unroll for store-stores chain.
16040 (prepare_initializers_chain_store_elim): New function.
16041 (prepare_initializers_chain): Hanlde store-store chain.
16042 (prepare_finalizers_chain, prepare_finalizers): New function.
16043 (tree_predictive_commoning_loop): Return integer value indicating
16044 if loop is unrolled or lcssa form is corrupted.
16045 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16047 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16049 * tree-predcom.c (initialize_root): Delete.
16050 (execute_pred_commoning_chain): Initialize root vars and replace
16051 reference of non-combined chain directly, rather than call above
16054 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16056 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16057 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16059 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16061 * tree-predcom.c (struct chain): New field init_seq.
16062 (release_chain): Release init_seq.
16063 (prepare_initializers_chain): Record intialization stmts in above
16065 (insert_init_seqs): New function.
16066 (tree_predictive_commoning_loop): Call insert_init_seqs.
16068 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16070 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16072 2017-07-28 Richard Biener <rguenther@suse.de>
16074 * match.pd: Remove superfluous :c.
16075 * genmatch.c (simplify::id): Add member.
16076 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16078 (current_id): New global.
16079 (dt_node::parent): Move from ...
16080 (dt_operand::parent): ... here. Add for_id member.
16081 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16082 (decision_tree::find_node): Relax order requirement when
16083 merging DT_TRUE nodes to ones inbetween the current simplify
16084 and the one we try to merge with. Add diagnostic whenever
16085 we need to enforce pattern order by not merging.
16086 (decision_tree::insert): Set current_id.
16087 (decision_tree::print_node): Dump parent node and for_id.
16088 (parser::last_id): Add member.
16089 (parser::push_simplify): Assign unique id.
16090 (parser::parser): Initialize last_id.
16092 2017-07-28 Martin Liska <mliska@suse.cz>
16095 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16096 gimple_build_debug_bind.
16098 2017-07-28 Richard Biener <rguenther@suse.de>
16100 PR tree-optimization/81502
16101 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16103 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16105 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16106 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16107 for BIT_FIELD_REF args.
16108 * fold-const.c (make_bit_field_ref): Likewise.
16109 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16111 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16114 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16115 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16116 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16117 Or it into SANITIZER_UNDEFINED.
16118 * ubsan.c: Include gimple-fold.h and varasm.h.
16119 (ubsan_expand_ptr_ifn): New function.
16120 (instrument_pointer_overflow): New function.
16121 (maybe_instrument_pointer_overflow): New function.
16122 (instrument_object_size): Formatting fix.
16123 (pass_ubsan::execute): Call instrument_pointer_overflow
16124 and maybe_instrument_pointer_overflow.
16125 * internal-fn.c (expand_UBSAN_PTR): New function.
16126 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16127 * sanitizer.def (__ubsan_handle_pointer_overflow,
16128 __ubsan_handle_pointer_overflow_abort): New builtins.
16129 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16130 * internal-fn.def (UBSAN_PTR): New internal function.
16131 * opts.c (sanitizer_opts): Add pointer-overflow.
16132 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16133 * fold-const.c (build_range_check): Compute pointer range check in
16134 integral type if pointer arithmetics would be needed. Formatting
16137 2017-07-28 Martin Liska <mliska@suse.cz>
16140 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16141 parameters that are of a variable-length.
16143 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16145 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16147 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16148 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16149 (CRT_CALL_STATIC_FUNCTION): Likewise.
16150 (ASM_DEFAULT_SPEC): New define.
16151 (ASM_SPEC32): Likewise.
16152 (ASM_SPEC64): Likewise.
16153 (ASM_SPEC_COMMON): Likewise.
16154 (ASM_SPEC): Likewise.
16155 (INVALID_64BIT): Likewise.
16156 (LINK_OS_DEFAULT_SPEC): Likewise.
16157 (LINK_OS_SPEC32): Likewise.
16158 (LINK_OS_SPEC64): Likewise.
16159 (POWERPC_LINUX): Likewise.
16160 (PTRDIFF_TYPE): Likewise.
16161 (RESTORE_FP_PREFIX): Likewise.
16162 (RESTORE_FP_SUFFIX): Likewise.
16163 (SAVE_FP_PREFIX): Likewise.
16164 (SAVE_FP_SUFFIX): Likewise.
16165 (SIZE_TYPE): Likewise.
16166 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16167 (TARGET_64BIT): Likewise.
16168 (TARGET_64BIT): Likewise.
16169 (TARGET_AIX): Likewise.
16170 (WCHAR_TYPE_SIZE): Likewise.
16171 (WCHAR_TYPE): Undefine.
16172 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16173 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16174 (CPP_OS_RTEMS_SPEC): Delete.
16175 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16176 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16178 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16180 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16182 PR tree-optimization/81578
16183 * tree-parloops.c (build_new_reduction): Bail out if
16184 reduction_code isn't one of the standard OpenMP reductions.
16185 Move the details printing after that decision.
16187 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16189 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16190 related to reload_in_progress.
16191 (splat_input_operand): Likewise.
16192 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16194 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16196 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16197 (TARGET_INSTANTIATE_DECLS): Likewise.
16198 (legitimate_indexed_address_p): Delete reload_in_progress code.
16199 (rs6000_debug_legitimate_address_p): Likewise.
16200 (rs6000_eliminate_indexed_memrefs): Likewise.
16201 (rs6000_emit_le_vsx_store): Likewise.
16202 (rs6000_emit_move_si_sf_subreg): Likewise.
16203 (rs6000_emit_move): Likewise.
16204 (register_to_reg_type): Likewise.
16205 (rs6000_pre_atomic_barrier): Likewise.
16206 (rs6000_machopic_legitimize_pic_address): Likewise.
16207 (rs6000_allocate_stack_temp): Likewise.
16208 (rs6000_address_for_fpconvert): Likewise.
16209 (rs6000_address_for_altivec): Likewise.
16210 (rs6000_secondary_memory_needed_rtx): Delete function.
16211 (rs6000_check_sdmode): Likewise.
16212 (rs6000_alloc_sdmode_stack_slot): Likewise.
16213 (rs6000_instantiate_decls): Likewise.
16214 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16215 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16216 Delete reload_in_progress.
16217 (*vec_reload_and_plus_<mptrsize>): Likewise.
16218 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16219 (vsx_div_v2di): Likewise.
16220 (vsx_udiv_v2di): Likewise.
16222 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16224 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16225 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16226 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16227 (rs6000_debug_reg_global): Delete print of LRA status.
16228 (rs6000_option_override_internal): Delete dead LRA related code.
16229 (rs6000_lra_p): Delete function.
16230 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16232 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16234 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16235 * config/riscv/rtems.h: New file.
16237 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16238 Sudakshina Das <sudi.das@arm.com>
16240 * config/aarch64/aarch64.md
16241 (define_split for and<mode>3nr_compare): Move
16242 non aarch64_logical_operand to a register.
16243 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16244 register immediate operand to a register.
16245 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16247 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16249 PR middle-end/81564
16250 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16252 2017-07-27 Richard Biener <rguenther@suse.de>
16254 PR tree-optimization/81573
16255 PR tree-optimization/81494
16256 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16257 multi defuse cycle case.
16259 2017-07-27 Richard Biener <rguenther@suse.de>
16261 PR tree-optimization/81571
16262 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16265 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16267 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16268 earlier and only if MASK_FPU is set. Adjust formatting.
16270 2017-07-27 Martin Liska <mliska@suse.cz>
16272 * opt-functions.awk: Add validation of value of Init.
16273 * optc-gen.awk: Pass new argument.
16275 2017-07-27 Martin Liska <mliska@suse.cz>
16277 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16278 Fix wrong condition.
16280 2017-07-27 Martin Liska <mliska@suse.cz>
16282 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16283 BBs and edges seen by autoFDO.
16285 2017-07-27 Richard Biener <rguenther@suse.de>
16287 PR tree-optimization/81502
16288 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16289 with incompatible but same sized type.
16290 (execute_update_addresses_taken): Likewise.
16292 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16294 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16295 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16297 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16300 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16301 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16302 Change s_operand to memory_operand.
16304 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16306 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16307 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16308 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16309 Emit instructions rather than returning an expression. Handle TFmode
16310 and KFmode by casting to TImode.
16311 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16312 (rs6000_emit_le_vsx_store): Likewise.
16313 * config/rs6000/vsx.md (VSX_TI): New iterator.
16314 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16315 (*vsx_le_undo_permute_<mode>): Likewise.
16316 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16317 emit the split sequence.
16318 (*vsx_le_perm_store_<mode>): Likewise.
16320 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16322 PR tree-optimization/81555
16323 PR tree-optimization/81556
16324 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16325 if true, force CHANGED for the recursive invocation.
16326 (reassociate_bb): Remember original length of ops array, pass
16327 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16329 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16330 attributes for noipa attribute. For naked attribute use
16331 lookup_attribute first before lookup_attribute_spec.
16332 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16334 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16335 for functions with noipa attribute.
16336 (cgraph_externally_visible_p): Return true for functions with noipa
16338 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16339 for functions with noipa attribute.
16340 * doc/extend.texi: Document noipa function attribute.
16341 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16342 also for functions with noipa attribute.
16343 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16345 2017-07-26 Andrew Pinski <apinski@cavium.com>
16347 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16348 vec_unalign_load_cost and vec_unalign_store_cost.
16350 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16352 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16353 -mvsx-small-integer option.
16354 (ISA_3_0_MASKS_IEEE): Likewise.
16355 (OTHER_VSX_VECTOR_MASKS): Likewise.
16356 (POWERPC_MASKS): Likewise.
16357 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16358 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16359 code, only testing for DImode being allowed in non-VSX floating
16361 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16362 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16364 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16365 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16366 TARGET_P8_VECTOR test.
16367 (rs6000_secondary_reload_simple_move): Likewise.
16368 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16369 since TARGET_P9_VECTOR was already tested.
16370 (rs6000_opt_masks): Remove -mvsx-small-integer.
16371 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16372 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16374 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16375 test for TARGET_VEXTRACTUB was used, and that uses
16377 (p9 extract splitter): Likewise.
16378 (vsx_extract_<mode>_di_p9): Likewise.
16379 (vsx_extract_<mode>_store_p9): Likewise.
16380 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16381 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16382 the elimination of TARGET_VSX_SMALL_INTEGER.
16383 (vsx_extract_<mode>_p8): Likewise.
16384 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16385 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16386 (vsx_set_<mode>_p9): Likewise.
16387 (vsx_set_v4sf_p9): Likewise.
16388 (vsx_set_v4sf_p9_zero): Likewise.
16389 (vsx_insert_extract_v4sf_p9): Likewise.
16390 (vsx_insert_extract_v4sf_p9_2): Likewise.
16391 * config/rs6000/rs6000.md (sign extend splitter): Change
16392 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16393 (floatsi<mode>2_lfiwax_mem): Likewise.
16394 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16395 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16396 since a test for TARGET_P9_VECTOR was used.
16397 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16398 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16399 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16400 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16401 TARGET_P8_VECTOR test.
16402 (fix_trunc<mode>si2_stfiwx): Likewise.
16403 (fix_trunc<mode>si2_internal): Likewise.
16404 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16405 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16407 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16408 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16409 TARGET_P8_VECTOR test.
16410 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16411 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16412 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16414 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16415 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16416 since a test for TARGET_P9_VECTOR was used.
16417 (splitter for loading small constants): Likewise.
16419 2017-07-26 Andrew Pinski <apinski@cavium.com>
16421 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16424 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16427 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16428 (fp_valid_at): Likewise.
16430 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16432 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16433 (qdf24xx_addrcost_table): Likewise.
16434 (cortexa57_tunings): Update to use generic_branch_cost.
16435 (cortexa72_tunings): Likewise.
16436 (cortexa73_tunings): Likewise.
16437 (qdf24xx_tunings): Likewise.
16439 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16441 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16442 (thunderx2t99_branch_cost): Likewise.
16443 (cortexa35_tunings): Update to use generic_branch_cost.
16444 (cortexa53_tunings): Likewise.
16445 (cortexa57_tunings): Likewise.
16446 (cortexa72_tunings): Likewise.
16447 (cortexa73_tunings): Likewise.
16448 (thunderx2t99_tunings): Likewise.
16450 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16452 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16453 (sparc_option_override): Honour MASK_FSMULD.
16454 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16455 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16456 * config/sparc/sparc.opt (mfsmuld): New option.
16457 * doc/invoke.texi (mfsmuld): Document option.
16459 2017-07-26 Marek Polacek <polacek@redhat.com>
16461 PR middle-end/70992
16462 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16464 2017-07-26 Richard Biener <rguenther@suse.de>
16466 * gimple-match-head.c (do_valueize): Return OP if valueize
16468 (get_def): New helper to get at the def stmt of a SSA name
16469 if valueize allows.
16470 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16471 do_valueize to get at the def stmt.
16472 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16474 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16476 PR middle-end/46932
16477 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16479 2017-07-26 Martin Liska <mliska@suse.cz>
16482 * function.c (expand_function_start): Make expansion of
16483 nonlocal_goto_save_area after parm_birth_insn.
16485 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16487 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16488 from all CPU target flags enable members.
16490 2017-07-26 Richard Biener <rguenther@suse.de>
16492 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16493 (decision_tree::gen): Make 'type' const.
16494 (write_predicate): Likewise.
16496 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16498 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16499 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16500 (rs6000_option_override_internal): Likewise.
16501 (rs6000_expand_vector_set): Likewise.
16502 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16503 (TARGET_UPPER_REGS_SF): Likewise.
16504 (TARGET_UPPER_REGS_DI): Likewise.
16505 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16506 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16507 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16508 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16509 (Splitters for DI constants in Altivec registers): Likewise.
16510 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16511 (vsx_set_v4sf_p9): Likewise.
16512 (vsx_set_v4sf_p9_zero): Likewise.
16513 (vsx_insert_extract_v4sf_p9): Likewise.
16514 (vsx_insert_extract_v4sf_p9_2): Likewise.
16516 2017-07-25 Carl Love <cel@us.ibm.com>
16518 * doc/extend.texi: Update the built-in documentation file for the
16519 existing built-in functions
16520 vector signed char vec_cnttz (vector signed char);
16521 vector unsigned char vec_cnttz (vector unsigned char);
16522 vector signed short vec_cnttz (vector signed short);
16523 vector unsigned short vec_cnttz (vector unsigned short);
16524 vector signed int vec_cnttz (vector signed int);
16525 vector unsigned int vec_cnttz (vector unsigned int);
16526 vector signed long long vec_cnttz (vector signed long long);
16527 vector unsigned long long vec_cnttz (vector unsigned long long);
16529 2017-07-25 Andrew Pinski <apinski@cavium.com>
16531 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16532 accesses where the use is for the first operand of a BIT_INSERT.
16534 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16537 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16538 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16540 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16542 * config/i386/gstabs.h: Delete.
16543 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16545 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16547 * config/i386/i386.c (ix86_decompose_address): Do not check for
16548 register RTX when looking at index_reg or base_reg.
16549 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16551 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16553 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16554 to update EH info here.
16556 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16558 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16560 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16562 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16564 2017-07-25 Torsten Duwe <duwe@suse.de>
16566 * common.opt: Introduce -fpatchable-function-entry
16567 command line option, and its variables function_entry_patch_area_size
16568 and function_entry_patch_area_start.
16569 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16570 including a two-value parser.
16571 * target.def (print_patchable_function_entry): New target hook.
16572 * targhooks.h (default_print_patchable_function_entry): New function.
16573 * targhooks.c (default_print_patchable_function_entry): Likewise.
16574 * toplev.c (process_options): Switch off IPA-RA if
16575 patchable function entries are being generated.
16576 * varasm.c (assemble_start_function): Look at the
16577 patchable-function-entry command line switch and current
16578 function attributes and maybe generate NOP instructions by
16579 calling the print_patchable_function_entry hook.
16580 * doc/extend.texi: Document patchable_function_entry attribute.
16581 * doc/invoke.texi: Document -fpatchable_function_entry
16582 command line option.
16583 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16585 * doc/tm.texi: Re-generate.
16587 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16590 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16591 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16593 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16595 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16597 2017-07-25 Richard Biener <rguenther@suse.de>
16599 PR tree-optimization/81455
16600 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16601 not walk in cycles when looking for guards.
16603 2017-07-25 Richard Biener <rguenther@suse.de>
16605 PR tree-optimization/81529
16606 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16607 when optimizing backedge uses.
16609 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16611 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16613 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16614 to dl_section_ref. On AIX, append an expression to subtract
16615 the size of the section length to dl_section_ref.
16617 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16619 * configure.ac: If any of the config.* scripts fail, exit 1.
16620 * configure: Regenerate.
16622 2017-07-25 Richard Biener <rguenther@suse.de>
16624 PR middle-end/81546
16625 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16626 of immediate uses, be more verbose on errors.
16628 2017-07-25 Richard Biener <rguenther@suse.de>
16630 PR tree-optimization/81510
16631 * tree-vect-loop.c (vect_is_simple_reduction): When the
16632 reduction stmt is not inside the loop bail out.
16634 2017-07-25 Richard Biener <rguenther@suse.de>
16636 PR tree-optimization/81303
16637 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16638 profitability check against LOOP_VINFO_NITERSM1.
16640 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16642 * domwalk.c (cmp_bb_postorder): Simplify.
16643 (sort_bbs_postorder): New function. Use it...
16644 (dom_walker::walk): ...here to optimize common cases.
16646 2017-07-25 Martin Liska <mliska@suse.cz>
16649 * ipa-visibility.c (function_and_variable_visibility): Make the
16650 redirection just on target that supports aliasing.
16651 Fix GNU coding style.
16653 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16656 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16658 * config/arm/rtems.h: Likewise.
16659 * config/bfin/rtems.h: Likewise.
16660 * config/i386/rtemself.h: Likewise.
16661 * config/lm32/rtems.h: Likewise.
16662 * config/m32c/rtems.h: Likewise.
16663 * config/m68k/rtemself.h: Likewise.
16664 * config/microblaze/rtems.h: Likewise.
16665 * config/mips/rtems.h: Likewise.
16666 * config/moxie/rtems.h: Likewise.
16667 * config/nios2/rtems.h: Likewise.
16668 * config/powerpcspe/rtems.h: Likewise.
16669 * config/rs6000/rtems.h: Likewise.
16670 * config/rtems.h: Likewise.
16671 * config/sh/rtems.h: Likewise.
16672 * config/sh/rtemself.h: Likewise.
16673 * config/sparc/rtemself.h: Likewise.
16675 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16678 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16679 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16680 * tree-ssa-structalias.c (alias_get_name): Same.
16682 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16685 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16686 instructions if no du chain is found.
16688 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16690 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16692 2017-07-25 Richard Biener <rguenther@suse.de>
16694 PR middle-end/81505
16695 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16698 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16700 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16701 upper-regs options.
16702 (ISA_2_7_MASKS_SERVER): Likewise.
16703 (ISA_3_0_MASKS_IEEE): Likewise.
16704 (OTHER_P8_VECTOR_MASKS): Likewise.
16705 (OTHER_VSX_VECTOR_MASKS): Likewise.
16706 (POWERPC_MASKS): Likewise.
16707 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16708 duplicate list of options.
16709 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16710 explicit -mupper-regs options.
16711 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16712 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16713 alias for -mupper-regs-df.
16714 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16715 (rs6000_init_hard_regno_mode_ok): Likewise.
16716 (rs6000_option_override_internal): Likewise.
16717 (rs6000_opt_masks): Likewise.
16718 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16719 options in terms of whether -mvsx or -mpower8-vector was used.
16720 (TARGET_UPPER_REGS_DI): Likewise.
16721 (TARGET_UPPER_REGS_SF): Likewise.
16722 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16723 -mupper-regs-* options.
16725 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16727 * passes.c (emergency_dump_function): Print some empty lines and a
16728 header before the RTL dump.
16730 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16732 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16734 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16737 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16738 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16740 2017-07-24 Carl Love <cel@us.ibm.com>
16742 * config/rs6000/rs6000-c.c: Add support for built-in functions
16743 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16744 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16745 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16746 vec_extract_fp_from_shortl): Add defines for the two builtins.
16747 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16748 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16750 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16751 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16752 * doc/extend.texi: Update the built-in documentation file for the
16753 new built-in function.
16755 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16758 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16760 * doc/generic.texi: Likewise.
16761 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16762 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16764 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16766 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16767 (aarch64_mls_elt_merge<mode>): Likewise.
16769 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16771 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16772 having __cxa_atexit.
16774 2017-07-23 Michael Collison <michael.collison@arm.com>
16776 * config/arm/arm.c (arm_option_override): Deprecate
16777 use of -mstructure-size-boundary.
16778 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16779 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16781 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16784 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16785 Reduce cost estimate for direct moves.
16787 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16790 * config/i386/i386.c (ix86_option_override_internal): Disable
16791 BMI, BMI2 and TBM instructions for -m16.
16793 2017-07-21 Carl Love <cel@us.ibm.com>
16795 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16796 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16797 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16798 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16799 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16800 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16801 VMULOSW): New enum "unspec" values.
16802 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16803 altivec_vmulosw): New patterns.
16804 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16805 VMULOSW): Add definitions.
16807 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16809 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16810 (qdf24xx): Likewise.
16811 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16812 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16813 (AARCH64_FL_V8_1): Renumber.
16814 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16815 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16816 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16817 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16818 rdma to feature modifiers list.
16820 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16822 PR middle-end/56727
16823 * ipa-visibility (function_and_variable_visibility): Convert
16824 recursive PLT call to direct call if appropriate.
16826 2017-07-21 Andrew Pinski <apinski@cavium.com>
16828 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16829 operand 1 to see if the types precision matches.
16830 * fold-const.c (operand_equal_p): Likewise.
16832 2017-07-21 Richard Biener <rguenther@suse.de>
16834 PR tree-optimization/81303
16835 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16836 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16837 (vect_peeling_hash_get_lowest_cost): Adjust.
16838 (vect_enhance_data_refs_alignment): Likewise. Use
16839 vect_get_peeling_costs_all_drs to compute the penalty for no
16840 peeling to match up costs.
16842 2017-07-21 Richard Biener <rguenther@suse.de>
16844 PR tree-optimization/81500
16845 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16846 we didn't identify a reduction path.
16848 2017-07-21 Tom de Vries <tom@codesourcery.com>
16849 Cesar Philippidis <cesar@codesourcery.com>
16851 PR gcov-profile/81442
16852 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16855 2017-07-21 Tom de Vries <tom@codesourcery.com>
16858 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16860 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16861 nvptx_override_options_after_change.
16863 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16865 * dwarf2out.c (output_file_names): Avoid double testing for
16866 dwarf_version >= 5.
16868 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16870 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16872 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16874 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16876 (try_crossjump_to_edge): Do not punt on partitioned functions.
16878 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16880 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16881 Put all BBs reachable only via paths crossing cold region to cold
16883 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16885 2016-07-21 Richard Biener <rguenther@suse.de>
16887 PR tree-optimization/81303
16888 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16889 into account prologue and epilogue iterations when raising
16890 min_profitable_iters to sth at least covering one vector iteration.
16892 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16894 * config/arm/arm.c (arm_test_cpu_arch_dat):
16897 2017-07-20 Nathan Sidwell <nathan@acm.org>
16899 Remove TYPE_METHODS.
16900 * tree.h (TYPE_METHODS): Delete.
16901 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16902 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16903 (dbxout_type_methods): Scan TYPE_FIELDS.
16904 (dbxout_type): Don't check TYPE_METHODS here.
16905 * function.c (use_register_for_decl): Always ignore register for
16906 class types when not optimizing.
16907 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16908 * tree.c (free_lang_data_in_type): Stitch out member functions and
16909 templates from TYPE_FIELDS.
16910 (build_distinct_type_copy, verify_type_variant,
16911 verify_type): Member fns are on TYPE_FIELDS.
16912 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16913 * tree-pretty-print.c (dump_generic_node): Likewise.
16915 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16918 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16919 V2TImode and V4TImode.
16920 (ix86_expand_vector_extract): Likewise.
16921 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16922 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16923 (ssescalarmode): Handle V4TImode and V2TImode.
16924 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16925 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16926 (VEXTRACTI128_MODE): New mode iterator.
16927 (splitter for *vec_extractv?ti first element): New.
16928 (VEC_INIT_MODE): New mode iterator.
16929 (vec_init<mode>): Consolidate 3 expanders into one using
16930 VEC_INIT_MODE mode iterator.
16932 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16934 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16935 non_spilled_static_chain_regno_p.
16937 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16939 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16941 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16943 * bb-reorder.c (connect_traces): Allow copying of blocks within
16946 2017-07-20 Richard Biener <rguenther@suse.de>
16948 * gimple.h (gimple_phi_result): Add gphi * overload.
16949 (gimple_phi_result_ptr): Likewise.
16950 (gimple_phi_arg): Likewise. Adjust index assert to only
16951 allow actual argument accesses rather than all slots available
16953 (gimple_phi_arg_def): Add gphi * overload.
16954 * tree-phinodes.c (make_phi_node): Initialize only actual
16956 (resize_phi_node): Clear memory not covered by old node,
16957 do not initialize excess argument slots.
16958 (reserve_phi_args_for_new_edge): Initialize new argument slot
16961 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16963 PR tree-optimization/81388
16965 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16967 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16968 by removing computation of may_be_zero.
16970 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16971 Tom de Vries <tom@codesourcery.com>
16973 PR middle-end/81030
16974 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16975 when gimple level profile disagrees with what RTL expander did.
16977 2017-07-20 Richard Biener <rguenther@suse.de>
16979 PR tree-optimization/61171
16980 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16981 (vect_analyze_stmt): Add slp instance parameter.
16982 (vectorizable_reduction): Likewise.
16983 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16984 (vect_is_simple_reduction): Deal with chains not detected
16985 as SLP reduction chain, specifically not properly associated
16986 chains containing a mix of plus/minus.
16987 (get_reduction_op): Remove.
16988 (get_initial_defs_for_reduction): Simplify, pass in whether
16989 this is a reduction chain, pass in the SLP node for the PHIs.
16990 (vect_create_epilog_for_reduction): Get the SLP instance as
16992 (vectorizable_reduction): Get the SLP instance as arg.
16993 During analysis remember the SLP node with the PHIs in the
16994 instance. Simplify getting at the vectorized reduction PHIs.
16995 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16996 through SLP instance.
16997 (vect_slp_analyze_operations): Likewise.
16998 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16999 (vect_transform_stmt): Likewise.
17001 2017-07-20 Tom de Vries <tom@codesourcery.com>
17003 PR tree-optimization/81489
17004 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17005 read of phi arg location to before loop that modifies phi.
17007 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17009 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17012 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17014 PR middle-end/81331
17015 * except.c (execute): Fix ordering issue.
17017 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17019 PR rtl-optimization/81423
17020 * combine.c (make_compound_operation_int): Don't try to optimize
17021 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17023 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17025 PR rtl-optimization/81423
17026 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17027 with a constant that is -1 in the truncated to mode.
17029 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17031 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17032 (determine_unlikely_bbs): ... here.
17033 * predict.h (propagate_unlikely_bbs_forward): Declare.
17034 * cfgexpand.c (pass_expand::execute): Use it.
17035 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17037 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17038 propagate_unlikely_bbs_forward.
17040 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17042 PR middle-end/81331
17043 * except.c (maybe_add_nop_after_section_switch): New function.
17046 2017-07-19 Tom de Vries <tom@codesourcery.com>
17048 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17050 2017-07-19 Tom de Vries <tom@codesourcery.com>
17052 * gimple.h (gimple_phi_arg): Make assert more strict.
17054 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17056 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17057 * config/rs6000/mmintrin.h: New file.
17058 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17060 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17062 PR tree-optimization/81346
17063 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17065 2017-07-19 Tom de Vries <tom@codesourcery.com>
17067 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17069 2017-07-19 Tom de Vries <tom@codesourcery.com>
17071 * config/nvptx/nvptx-modes.def: Add V2DImode.
17072 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17073 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17074 (nvptx_output_mov_insn): Handle lack of mov.b128.
17075 (nvptx_print_operand): Handle 'H' and 'L' codes.
17076 (nvptx_vector_mode_supported): Allow V2DImode.
17077 (nvptx_preferred_simd_mode): New function.
17078 (nvptx_data_alignment): New function.
17079 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17080 nvptx_preferred_simd_mode.
17081 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17083 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17085 2017-07-19 Tom de Vries <tom@codesourcery.com>
17087 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17088 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17089 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17090 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17091 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17092 (mov<VECIM>_insn): New define_insn.
17093 (define_expand "mov<VECIM>): New define_expand.
17095 2017-07-19 Tom de Vries <tom@codesourcery.com>
17097 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17099 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17101 PR tree-optimization/81346
17102 * fold-const.h (fold_div_compare, range_check_type): Declare.
17103 * fold-const.c (range_check_type): New function.
17104 (build_range_check): Use range_check_type.
17105 (fold_div_compare): No longer static, rewritten into
17106 a match.pd helper function.
17107 (fold_comparison): Don't call fold_div_compare here.
17108 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17109 as helper function.
17111 2017-07-19 Nathan Sidwell <nathan@acm.org>
17113 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17114 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17115 * tree.c (find_decls_types_r, verify_type): Use
17116 TYPE_{MIN,MAX}_VALUE_RAW.
17117 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17118 (hash_tree): Likewise.
17119 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17121 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17124 2017-07-18 Tom de Vries <tom@codesourcery.com>
17126 PR middle-end/81464
17127 * omp-expand.c (expand_omp_for_static_chunk): Handle
17128 equal-argument loop exit phi.
17130 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17133 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17134 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17135 operand 2 predicate.
17136 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17137 operand 2 predicate.
17138 (ror,rol -> rorx splitters): Use const_int_operand as
17139 operand 2 predicate.
17141 2017-06-18 Richard Biener <rguenther@suse.de>
17143 PR tree-optimization/81410
17144 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17145 the gap in the ! slp_perm SLP case after each group.
17147 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17149 PR middle-end/81463
17150 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17153 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17155 PR middle-end/81462
17156 * predict.c (set_even_probabilities): Cleanup; do not affect
17157 probabilities that are already known.
17158 (combine_predictions_for_bb): Call even when count is set.
17160 2017-07-18 Nathan Sidwell <nathan@acm.org>
17162 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17165 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17168 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17169 optimization for loop niter analysis.
17171 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17174 * config/avr/avr.c (avr_optimize_casesi): Don't use
17175 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17177 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17179 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17180 body_cost_vec from _vect_peel_extended_info.
17181 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17182 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17185 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17187 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17189 2017-07-18 Richard Biener <rguenther@suse.de>
17191 PR tree-optimization/80620
17192 PR tree-optimization/81403
17193 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17194 info when re-using a VN table entry.
17196 2017-07-18 Richard Biener <rguenther@suse.de>
17198 PR tree-optimization/81418
17199 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17200 vectype_in. Verify that with lane-reducing reduction operations
17201 we have a single def-use cycle.
17203 2017-07-17 Carl Love <cel@us.ibm.com>
17205 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17207 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17208 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17209 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17210 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17211 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17212 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17213 VMULOSW): New enum "unspec" values.
17214 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17215 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17216 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17217 altivec_vmulosw): New patterns.
17218 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17219 VMULOSW): Add definitions.
17220 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17222 * config/alpha/alpha.c: Include predict.h.
17224 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17226 * tree-vrp.c (compare_assert_loc): Fix comparison function
17227 to return predictable results.
17229 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17231 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17233 (subdi3): Likewise.
17234 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17235 * doc/invoke.texi (mexpand-adddi): Update text.
17237 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17239 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17240 that also clobbers the CC register. The old expand code is moved
17242 (*arc_clzsi2): ... here.
17243 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17244 the CC register. The old expand code is moved to ...
17245 (arc_ctzsi2): ... here.
17247 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17249 * config/arc/arc.opt (mindexed-loads): Use initial value
17250 TARGET_INDEXED_LOADS_DEFAULT.
17251 (mauto-modify-reg): Use initial value
17252 TARGET_AUTO_MODIFY_REG_DEFAULT.
17253 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17254 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17255 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17256 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17258 2017-07-17 Martin Liska <mliska@suse.cz>
17261 * opts.c (finish_options): Do not allow -fgnu-tm
17262 w/ -fsanitize={kernel-,}address. Say sorry.
17264 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17267 * tree-loop-distribution.c (classify_partition): Only assert on
17268 numer of iterations.
17269 (merge_dep_scc_partitions): Delete prameter. Update function call.
17270 (distribute_loop): Remove code handling loop with unknown niters.
17271 (pass_loop_distribution::execute): Skip loop with unknown niters.
17273 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17276 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17277 function sort_partitions_by_post_order.
17279 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17281 PR tree-optimization/81374
17282 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17283 the max index of basic blocks, rather than number of basic blocks.
17285 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17287 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17289 (arc_legitimate_pic_operand_p): Likewise.
17290 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17292 (arc_needs_pcl_p): Likewise.
17293 (arc_legitimate_pc_offset_p): Likewise.
17294 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17295 function is also used in constrains.md.
17296 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17297 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17298 PLUS. Only return true/false in known cases, otherwise assert.
17299 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17300 is already called in arc_legitimate_constant_p.
17301 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17303 (LEGITIMATE_PIC_OPERAND_P): Use
17304 arc_raw_symbolic_reference_mentioned_p function.
17305 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17310 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17311 Andrew Burgess <andrew.burgess@embecosm.com>
17313 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17314 (arc_return_address_register): New function.
17315 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17316 (arc_handle_fndecl_attribute): Add naked attribute.
17317 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17318 (TARGET_WARN_FUNC_RETURN): Likewise.
17319 (arc_allocate_stack_slots_for_args): New function.
17320 (arc_warn_func_return): Likewise.
17321 (machine_function): Change type fn_type.
17322 (arc_compute_function_type): Consider new naked function type,
17323 change function return type.
17324 (arc_must_save_register): Adapt to handle new
17325 arc_compute_function_type's return type.
17326 (arc_expand_prologue): Likewise.
17327 (arc_expand_epilogue): Likewise.
17328 (arc_return_address_regs): Delete.
17329 (arc_return_address_register): New function.
17330 (arc_epilogue_uses): Use above function.
17331 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17332 (arc_function_type): Change encoding, add naked type.
17333 (ARC_INTERRUPT_P): Change to handle the new encoding.
17334 (ARC_FAST_INTERRUPT_P): Likewise.
17335 (ARC_NORMAL_P): Define.
17336 (ARC_NAKED_P): Likewise.
17337 (arc_compute_function_type): Delete prototype.
17338 * config/arc/arc.md (in_ret_delay_slot): Use
17339 arc_return_address_register function.
17340 (simple_return): Likewise.
17341 (p_return_i): Likewise.
17343 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17345 PR tree-optimization/81428
17346 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17347 can't be built for those types.
17349 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17351 Remove stuff dead since r239246.
17353 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17354 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17355 (avr_inform_devices): Remove dead stuff.
17357 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17359 * config/arm/arm_neon.h: Fix softp typo.
17361 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17363 PR tree-optimization/81365
17364 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17365 aggregate moves onto bb predecessor edges, make sure there are no
17366 loads that could alias the lhs in between the start of bb and the
17369 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17372 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17373 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17374 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17376 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17378 PR tree-optimization/81396
17379 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17380 (init_symbolic_number): Initialize it to 1.
17381 (perform_symbolic_merge): Add n_ops from both operands into the new
17383 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17384 without base_addr as useless if they need more than one operation.
17385 (bswap_replace): Handle !bswap case for NULL base_addr.
17387 2017-07-17 Tom de Vries <tom@codesourcery.com>
17390 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17393 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17395 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17396 conditional builtin define __FIX_LEON3FT_B2BST.
17398 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17400 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17401 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17404 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17406 PR rtl-optimization/81424
17407 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17408 to remove potential trapping from operands if -fnon-call-exceptions.
17410 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17412 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17413 profile_proability for scalling.
17414 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17416 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17418 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17420 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17422 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17423 fixpoint arithmetics.
17425 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17427 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17428 fixpoint arithmetics.
17430 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17432 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17433 fixpoint arithmetics.
17435 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17437 * profile-count.h (profile_probability::from_reg_br_prob_note,
17438 profile_probability::to_reg_br_prob_note): New functions.
17439 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17440 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17441 * predict.c (probability_reliable_p): Update.
17442 (edge_probability_reliable_p): Update.
17443 (br_prob_note_reliable_p): Update.
17444 (invert_br_probabilities): Update.
17445 (add_reg_br_prob_note): New function.
17446 (combine_predictions_for_insn): Update.
17447 * asan.c (asan_clear_shadow): Update.
17448 * cfgbuild.c (compute_outgoing_frequencies): Update.
17449 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17450 (update_br_prob_note): Update.
17451 (rtl_verify_edges): Update.
17452 (purge_dead_edges): Update.
17453 (fixup_reorder_chain): Update.
17454 * emit-rtl.c (try_split): Update.
17455 * ifcvt.c (cond_exec_process_insns): Update.
17456 (cond_exec_process_if_block): Update.
17457 (dead_or_predicable): Update.
17458 * internal-fn.c (expand_addsub_overflow): Update.
17459 (expand_neg_overflow): Update.
17460 (expand_mul_overflow): Update.
17461 * loop-doloop.c (doloop_modify): Update.
17462 * loop-unroll.c (compare_and_jump_seq): Update.
17463 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17464 * predict.h: Update.
17465 * reorg.c (mostly_true_jump): Update.
17467 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17468 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17469 * config/arc/arc.c: (emit_unlikely_jump): Update.
17470 * config/arm/arm.c: (emit_unlikely_jump): Update.
17471 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17472 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17473 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17474 (ix86_print_operand): Update.
17475 (ix86_split_fp_branch): Update.
17476 (predict_jump): Update.
17477 * config/ia64/ia64.c (ia64_print_operand): Update.
17478 * config/mmix/mmix.c (mmix_print_operand): Update.
17479 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17480 (rs6000_expand_split_stack_prologue): Update.
17481 * config/rs6000/rs6000.c: Update.
17482 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17483 (s390_expand_vec_movstr): Update.
17484 (s390_expand_cs_tdsi): Update.
17485 (s390_expand_split_stack_prologue): Update.
17486 * config/sh/sh.c (sh_print_operand): Update.
17487 (expand_cbranchsi4): Update.
17488 (expand_cbranchdi4): Update.
17489 * config/sparc/sparc.c (output_v9branch): Update.
17490 * config/spu/spu.c (get_branch_target): Update.
17491 (ea_load_store_inline): Update.
17492 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17493 * config/tilepro/tilepro.c: Update.
17495 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17497 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17498 (gimplify_save_expr): Likewise.
17500 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17502 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17504 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17506 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17508 * timevar.def (TV_IPA_FNSUMMARY): Define.
17510 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17512 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17513 to back store errata sensitive sequence from being generated.
17514 (sqrtdf2_fix): Likewise.
17516 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17518 * tree-ssa-threadupdate.c (compute_path_counts,
17519 update_joiner_offpath_counts): Use profile_probability.
17521 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17524 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17526 * config/arm/arm-c.c (arm_cpu_builtins): Define
17527 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17529 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17531 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17532 array entries to represent __ieee128 versions of the
17533 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17534 scalar_extract_sig, and scalar_insert_exp built-in functions.
17535 (altivec_resolve_overloaded_builtin): Add special case handling
17536 for the __builtin_scalar_insert_exp function, as represented by
17537 the P9V_BUILTIN_VEC_VSIEDP constant.
17538 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17539 exponent support for __ieee128 argument.
17540 (VSESQP): Add scalar extract signature support for __ieee128
17542 (VSTDCNQP): Add scalar test negative support for __ieee128
17544 (VSIEQP): Add scalar insert exponent support for __int128 argument
17545 with __ieee128 result.
17546 (VSIEQPF): Add scalar insert exponent support for __ieee128
17547 argument with __ieee128 result.
17548 (VSTDCQP): Add scalar test data class support for __ieee128
17550 (VSTDCNQP): Add overload support for scalar test negative with
17551 __ieee128 argument.
17552 (VSTDCQP): Add overload support for scalar test data class
17553 __ieee128 argument.
17554 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17555 UNSPEC_VSX_SXSIGDP.
17556 (UNSPEC_VSX_SIEXPQP): New constant.
17557 (xsxexpqp): New insn for VSX scalar extract exponent quad
17559 (xsxsigqp): New insn for VSX scalar extract significand quad
17561 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17562 precision with floating point argument.
17563 (xststdcqp): New expand for VSX scalar test data class quad
17565 (xststdcnegqp): New expand for VSX scalar test negative quad
17567 (xststdcqp): New insn to match expansions for VSX scalar test data
17568 class quad precision and VSX scalar test negative quad precision.
17569 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17570 special case operand checking to enforce that second operand of
17571 VSX scalar test data class with quad precision argument is a 7-bit
17573 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17574 prototypes and descriptions of __ieee128 versions of
17575 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17576 scalar_test_data_class, and scalar_test_neg built-in functions.
17578 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17580 PR tree-optimization/81162
17581 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17582 replace a negate with an add.
17584 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17586 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17588 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17590 * config/arm/arm-c.c (arm_cpu_builtins): Define
17591 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17593 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17595 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17596 (armv8-r): Set ARM Cortex-R52 as default CPU.
17597 * config/arm/arm-tables.opt: Regenerate.
17598 * config/arm/arm-tune.md: Regenerate.
17599 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17601 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17602 extension for -mcpu=cortex-r52.
17604 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17606 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17607 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17608 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17609 (fp-armv8): Define it as FP_ARMv8 only.
17610 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17611 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17613 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17614 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17615 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17616 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17617 than TARGET_FPU_ARMV8.
17618 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17619 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17620 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17622 * config/arm/neon.md (neon_vrint): Likewise.
17623 (neon_vcvt): Likewise.
17624 (neon_<fmaxmin_op><mode>): Likewise.
17625 (<fmaxmin><mode>3): Likewise.
17626 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17627 * config/arm/predicates.md (arm_cond_move_operator): Check against
17628 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17630 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17632 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17633 to top of function.
17635 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17637 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17638 loop in comment with memset.
17640 2017-07-14 Martin Liska <mliska@suse.cz>
17642 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17643 * dwarf2out.c (is_java): Remove the function.
17644 (output_pubname): Remove usage of the function.
17645 (lower_bound_default): Remove usage of DW_LANG_Java.
17646 (gen_compile_unit_die): Likewise.
17647 * gcc.c: Remove compiler defaults for .java and .zip files.
17648 * gimple-expr.c (remove_suffix): Change as there's no longer
17649 extension than 4-letter one.
17650 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17651 (gimplify_save_expr): Likewise.
17652 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17653 as it's possible even for other languages than Java.
17654 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17655 * lto-opts.c (lto_write_options): Remove reference to Java.
17656 * opts.c (strip_off_ending): Update file extension handling.
17657 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17658 * tree-eh.c (lower_resx): Likewise.
17659 * tree.c (free_lang_data_in_type): Remove dead code.
17660 (find_decls_types_r): Likewise.
17661 (build_common_builtin_nodes): Remove Java from a comment.
17662 (verify_type): Remove dead code.
17663 * varasm.c (assemble_external): Remove Java from a comment.
17665 2017-07-14 Martin Liska <mliska@suse.cz>
17667 * opts.c (finish_options): Add quotes.
17668 (common_handle_option): Likewise.
17670 2017-07-14 Martin Liska <mliska@suse.cz>
17672 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17673 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17674 Remove N_SO_PASCAL.
17675 * dwarf2out.c (lower_bound_default): Do not handle
17677 (gen_compile_unit_die): Likewise.
17678 * gcc.c: Remove default extension binding for GNU Pascal.
17679 * stmt.c: Remove Pascal language from a comment.
17680 * xcoffout.c: Likewise.
17682 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17685 * diagnostic-show-locus.c (fixit_cmp): New function.
17686 (layout::layout): Sort m_fixit_hints.
17687 (column_range::column_range): Assert that the values are valid.
17688 (struct char_span): New struct.
17689 (correction::overwrite): New method.
17690 (struct source_line): New struct.
17691 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17692 calls in terms of classes source_line and char_span, and
17693 correction::overwrite.
17694 (selftest::test_overlapped_fixit_printing_2): New function.
17695 (selftest::diagnostic_show_locus_c_tests): Call it.
17697 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17699 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17700 early if there is no lhs.
17702 2017-07-13 Martin Liska <mliska@suse.cz>
17704 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17705 (gen_reference_type_die): Likewise.
17706 * stor-layout.c: Remove Pascal-related comment.
17708 2017-07-13 Martin Liska <mliska@suse.cz>
17710 * opts.c (finish_options): Add quotes to error messages.
17711 (parse_sanitizer_options): Likewise.
17713 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17715 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17717 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17719 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17721 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17723 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17725 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17727 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17730 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17731 provides the hardware capability bits, define the macro
17732 __BUILTIN_CPU_SUPPORTS__.
17733 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17734 if GLIBC does not provide the hardware capability bits. Add a
17735 gcc_unreachable call if the built-in cpu function is neither
17736 __builtin_cpu_is nor __builtin_cpu_supports.
17737 (rs6000_get_function_versions_dispatcher): Change the warning
17738 that an old GLIBC is used which does not export the capability
17739 bits to be an error.
17740 * doc/extend.texi (target_clones attribute): Document the
17741 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17742 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17743 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17744 the macros defined by GCC if the newer GLIBC is available.
17746 2017-07-12 Jeff Law <law@redhat.com>
17748 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17749 remaining includes slightly.
17750 * config/riscv/riscv-builtins.c: Include profile-count.h.
17752 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17755 * config/avr/avr.c (avr_set_current_function): In diagnostic
17756 messages: Quote keywords and (parts of) identifiers.
17757 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17760 2017-07-12 Carl Love <cel@us.ibm.com>
17762 * config/rs6000/rs6000-c.c: Add support for built-in functions
17763 vector bool char vec_revb (vector bool char);
17764 vector bool short vec_revb (vector short char);
17765 vector bool int vec_revb (vector bool int);
17766 vector bool long long vec_revb (vector bool long long);
17767 * doc/extend.texi: Update the built-in documentation file for the
17768 new built-in functions.
17770 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17772 * config/s390/s390.md: Remove movcc splitter.
17774 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17776 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17777 load/store on condition.
17779 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17782 * config/avr/avr.c (avr_encode_section_info)
17783 [progmem && !TREE_READONLY]: Error if progmem object needs
17786 2017-07-11 Michael Collison <michael.collison@arm.com>
17788 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17791 2017-07-11 Carl Love <cel@us.ibm.com>
17793 * config/rs6000/rs6000-c.c: Add support for builtins
17794 vector unsigned int vec_parity_lsbb (vector signed int);
17795 vector unsigned int vec_parity_lsbb (vector unsigned int);
17796 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17797 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17798 vector unsigned long long vec_parity_lsbb (vector signed long long);
17799 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17800 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17801 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17802 * doc/extend.texi: Update the built-in documentation file for the
17803 new built-in functions.
17805 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17807 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17808 (layout::m_primary_loc): New field.
17809 (layout::layout): Initialize new field. Move location filtering
17810 logic from here to...
17811 (layout::maybe_add_location_range): ...this new method. Add
17812 support for filtering to just the lines already specified by other
17814 (layout::will_show_line_p): New method.
17815 (gcc_rich_location::add_location_if_nearby): New method.
17816 (selftest::test_add_location_if_nearby): New test function.
17817 (selftest::diagnostic_show_locus_c_tests): Call it.
17818 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17821 2017-07-11 Tom de Vries <tom@codesourcery.com>
17823 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17824 (bb_first_real_insn): New function.
17825 (nvptx_single): Add extra initialization of broadcasted condition
17828 2017-07-11 Nathan Sidwell <nathan@acm.org>
17830 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17832 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17834 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17835 Binutils doc as TEXI will mess them up.
17836 * doc/invoke.texi (AVR Options): Same here.
17838 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17840 * config/sparc/sparc.opt (mfix-ut700): New option.
17841 (mfix-gr712rc): Likewise.
17842 (sparc_fix_b2bst): New variable.
17843 * doc/invoke.texi (SPARC options): Document them.
17844 (ARM options): Fix warnings.
17845 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17846 instructions to prevent sequences that can trigger the store-store
17847 errata for certain LEON3FT processors.
17848 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17849 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17850 * config/sparc/sparc.md (fix_b2bst): New attribute.
17851 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17853 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17856 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17858 (*rsqrtsf2_sse): Ditto.
17860 (div<mode>3): Macroize insn from divdf3 and divsf3
17861 using MODEF mode iterator.
17863 2017-07-10 Martin Sebor <msebor@redhat.com>
17865 PR tree-optimization/80397
17866 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17867 instead of testing for equality to INTEGER_TYPE.
17869 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17871 * config.gcc: Remove uclibc from arc target spec.
17873 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17875 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17877 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17880 * lto-wrapper.c (remove_option): New function.
17881 (merge_and_complain): Merge PIC/PIE options more realistically.
17883 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17885 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17889 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17890 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17891 * config.in: Regenerate.
17892 * configure: Regenerate.
17893 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17894 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17895 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17896 (TARGET_GASISR_PROLOGUES): ...target mask.
17897 * common/config/avr/avr-common.c
17898 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17899 Set -mgas-isr-prologues.
17900 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17901 INSERT_PASS_BEFORE for it.
17902 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17903 * config/avr/avr.c (avr_option_override)
17904 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17905 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17906 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17907 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17908 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17910 (make_avr_pass_pre_proep): New function.
17911 (emit_push_sfr) <treg>: Add argument to function and use it
17912 instead of TMP_REG.
17913 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17914 and set machine->gasisr.yes.
17915 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17916 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17917 __gcc_isr.n_pushed to .L__stack_usage.
17918 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17919 (avr_asm_final_postscan_insn): ...this new static function.
17920 * config/avr/avr.h (machine_function)
17921 <is_no_gccisr, use_L__stack_usage>: New fields.
17922 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17923 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17924 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17925 (gasisr, *gasisr): New expander and insn.
17926 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17927 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17928 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17930 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17932 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17935 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17937 Move jump-tables out of .text again.
17940 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17941 (ASM_OUTPUT_ADDR_VEC): New function.
17942 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17943 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17944 INSN_ADDRESSes as asm comment.
17945 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17946 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17947 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17948 * config/avr/avr.md (*tablejump): Adjust comment.
17949 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17950 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17952 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17953 (avr_output_addr_vec): New proto.
17954 (avr_log_t) <insn_addresses>: New field.
17956 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17959 * config/i386/i386.c (ix86_function_arg_advance): Set
17960 outgoing_args_on_stack to true if there are outgoing arguments
17962 (ix86_function_arg): Likewise.
17963 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17964 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17965 * config/i386/i386.h (machine_function): Add
17966 outgoing_args_on_stack.
17968 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17970 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17971 supporting pthreds.
17972 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17974 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17976 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17977 (REAL_H): Remove $(MACHMODE_H).
17978 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17980 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17981 $(MACHMODE_H) and double-int.h.
17982 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17984 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17987 2017-07-07 Andrew Pinski <apinski@cavium.com>
17989 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17990 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17992 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17994 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17995 Add warning if GCC was not configured to link against a GLIBC that
17996 exports the hardware capability bits.
17997 (make_resolver_func): Make resolver function private and not a
17998 COMDAT function. Create the name with clone_function_name instead
17999 of make_unique_name.
18002 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18003 correct operand in doing the split.
18005 2017-07-07 Carl Love <cel@us.ibm.com>
18007 * config/rs6000/rs6000-c: Add support for built-in function
18008 vector unsigned short vec_pack_to_short_fp32 (vector float,
18010 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18011 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18012 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18013 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18014 (convert_4f32_8i16): Add define_expand.
18015 * doc/extend.texi: Update the built-in documentation file for the
18016 new built-in function.
18018 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18020 * config/sparc/m8.md: New file.
18021 * config/sparc/sparc.md: Include m8.md.
18023 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18025 * config/sparc/sparc.opt: New option -mvis4b.
18026 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18027 (sparc_option_override): Handle VIS4B.
18028 (enum sparc_builtins): Define
18029 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18030 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18031 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18032 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18033 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18034 (check_constant_argument): New function.
18035 (sparc_vis_init_builtins): Define builtins
18036 __builtin_vis_dictunpack{8,16,32},
18037 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18038 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18039 __builtin_vis_fpcmpde{8,16,32}shl and
18040 __builtin_vis_fpcmpur{8,16,32}shl.
18041 (sparc_expand_builtin): Check that the constant operands to
18042 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18043 constant and in range.
18044 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18046 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18047 (SPARC_IMM5_P): Likewise.
18048 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18049 (enabled): Handle vis4b.
18050 (UNSPEC_DICTUNPACK): New unspec.
18051 (UNSPEC_FPCMPSHL): Likewise.
18052 (UNSPEC_FPUCMPSHL): Likewise.
18053 (UNSPEC_FPCMPDESHL): Likewise.
18054 (UNSPEC_FPCMPURSHL): Likewise.
18055 (cpu_feature): New CPU feature `vis4b'.
18056 (dictunpack{8,16,32}): New insns.
18057 (FPCSMODE): New mode iterator.
18058 (fpcscond): New code iterator.
18059 (fpcsucond): Likewise.
18060 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18061 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18062 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18063 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18064 * config/sparc/constraints.md: Define constraints `q' for unsigned
18065 2-bit integer constants and `t' for unsigned 5-bit integer
18067 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18069 (imm5_operand_dictunpack16): Likewise.
18070 (imm5_operand_dictunpack32): Likewise.
18071 (imm2_operand): Likewise.
18072 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18073 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18074 ditunpack* and fpcmp*shl builtins.
18076 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18078 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18079 * config.in: Add HAVE_AS_SPARC6 define.
18080 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18082 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18084 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18085 (CPP_CPU_SPEC): Handle m8.
18086 (ASM_CPU_SPEC): Likewise.
18087 * config/sparc/sparc-opts.h (enum processor_type): Add
18089 * config/sparc/sparc.c (m8_costs): New struct.
18090 (sparc_option_override): Handle TARGET_CPU_m8.
18091 (sparc32_initialize_trampoline): Likewise.
18092 (sparc64_initialize_trampoline): Likewise.
18093 (sparc_issue_rate): Likewise.
18094 (sparc_register_move_cost): Likewise.
18095 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18096 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18097 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18098 (CPP_CPU_SPEC): Handle M8.
18099 (ASM_CPU_SPEC): Likewise.
18100 (AS_M8_FLAG): Define.
18101 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18102 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18103 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18105 * configure: Regenerate.
18106 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18109 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18111 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18113 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18114 ("*movdi_insn_sp32"): Do not set v3pipe.
18115 ("*movsi_insn"): Likewise.
18116 ("*movdi_insn_sp64"): Likewise.
18117 ("*movsf_insn"): Likewise.
18118 ("*movdf_insn_sp32"): Likewise.
18119 ("*movdf_insn_sp64"): Likewise.
18120 ("*zero_extendsidi2_insn_sp64"): Likewise.
18121 ("*sign_extendsidi2_insn"): Likewise.
18122 ("*mov<VM32:mode>_insn"): Likewise.
18123 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18124 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18125 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18126 ("<vlop:code><VL:mode>3"): Likewise.
18127 ("*not_<vlop:code><VL:mode>3"): Likewise.
18128 ("*nand<VL:mode>_vis"): Likewise.
18129 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18130 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18131 ("one_cmpl<VL:mode>2"): Likewise.
18132 ("faligndata<VM64:mode>_vis"): Likewise.
18133 ("alignaddrsi_vis"): Likewise.
18134 ("alignaddrdi_vis"): Likweise.
18135 ("alignaddrlsi_vis"): Likewise.
18136 ("alignaddrldi_vis"): Likewise.
18137 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18138 ("bmaskdi_vis"): Likewise.
18139 ("bmasksi_vis"): Likewise.
18140 ("bshuffle<VM64:mode>_vis"): Likewise.
18141 ("cmask8<P:mode>_vis"): Likewise.
18142 ("cmask16<P:mode>_vis"): Likewise.
18143 ("cmask32<P:mode>_vis"): Likewise.
18144 ("pdistn<P:mode>_vis"): Likewise.
18145 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18147 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18149 * config/sparc/sparc.md ("subtype"): New insn attribute.
18150 ("*wrgsr_sp64"): Set insn subtype.
18151 ("*rdgsr_sp64"): Likewise.
18152 ("alignaddrsi_vis"): Likewise.
18153 ("alignaddrdi_vis"): Likewise.
18154 ("alignaddrlsi_vis"): Likewise.
18155 ("alignaddrldi_vis"): Likewise.
18156 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18157 ("fexpand_vis"): Likewise.
18158 ("fpmerge_vis"): Likewise.
18159 ("faligndata<VM64:mode>_vis"): Likewise.
18160 ("bshuffle<VM64:mode>_vis"): Likewise.
18161 ("cmask8<P:mode>_vis"): Likewise.
18162 ("cmask16<P:mode>_vis"): Likewise.
18163 ("cmask32<P:mode>_vis"): Likewise.
18164 ("fchksm16_vis"): Likewise.
18165 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18166 ("fmean16_vis"): Likewise.
18167 ("fp<plusminus_insn>64_vis"): Likewise.
18168 ("<plusminus_insn>v8qi3"): Likewise.
18169 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18170 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18171 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18172 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18173 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18174 ("*movqi_insn"): Likewise.
18175 ("*movhi_insn"): Likewise.
18176 ("*movsi_insn"): Likewise.
18177 ("movsi_pic_gotdata_op"): Likewise.
18178 ("*movdi_insn_sp32"): Likewise.
18179 ("*movdi_insn_sp64"): Likewise.
18180 ("movdi_pic_gotdata_op"): Likewise.
18181 ("*movsf_insn"): Likewise.
18182 ("*movdf_insn_sp32"): Likewise.
18183 ("*movdf_insn_sp64"): Likewise.
18184 ("*zero_extendhisi2_insn"): Likewise.
18185 ("*zero_extendqihi2_insn"): Likewise.
18186 ("*zero_extendqisi2_insn"): Likewise.
18187 ("*zero_extendqidi2_insn"): Likewise.
18188 ("*zero_extendhidi2_insn"): Likewise.
18189 ("*zero_extendsidi2_insn_sp64"): Likewise.
18190 ("ldfsr"): Likewise.
18191 ("prefetch_64"): Likewise.
18192 ("prefetch_32"): Likewise.
18193 ("tie_ld32"): Likewise.
18194 ("tie_ld64"): Likewise.
18195 ("*tldo_ldub_sp32"): Likewise.
18196 ("*tldo_ldub1_sp32"): Likewise.
18197 ("*tldo_ldub2_sp32"): Likewise.
18198 ("*tldo_ldub_sp64"): Likewise.
18199 ("*tldo_ldub1_sp64"): Likewise.
18200 ("*tldo_ldub2_sp64"): Likewise.
18201 ("*tldo_ldub3_sp64"): Likewise.
18202 ("*tldo_lduh_sp32"): Likewise.
18203 ("*tldo_lduh1_sp32"): Likewise.
18204 ("*tldo_lduh_sp64"): Likewise.
18205 ("*tldo_lduh1_sp64"): Likewise.
18206 ("*tldo_lduh2_sp64"): Likewise.
18207 ("*tldo_lduw_sp32"): Likewise.
18208 ("*tldo_lduw_sp64"): Likewise.
18209 ("*tldo_lduw1_sp64"): Likewise.
18210 ("*tldo_ldx_sp64"): Likewise.
18211 ("*mov<VM32:mode>_insn"): Likewise.
18212 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18213 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18215 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18217 * config/sparc/sparc.md ("type"): New insn type viscmp.
18218 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18220 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18221 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18222 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18223 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18225 ("n7_vis_logical_11cycle"): Likewise.
18226 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18227 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18228 * config/sparc/niagara.md ("niag_vis"): Likewise.
18229 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18230 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18232 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18234 * config/sparc/sparc.md: New instruction type `bmask'.
18235 (bmaskdi_vis): Use the `bmask' type.
18236 (bmasksi_vis): Likewise.
18237 * config/sparc/ultra3.md (us3_array): Likewise.
18238 * config/sparc/niagara7.md (n7_array): Likewise.
18239 * config/sparc/niagara4.md (n4_array): Likewise.
18240 * config/sparc/niagara2.md (niag2_vis): Likewise.
18241 (niag3_vis): Likewise.
18242 * config/sparc/niagara.md (niag_vis): Likewise.
18244 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18246 * ipa-comdats.c: Remove optimize check from gate.
18247 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18248 for functions not optimized.
18249 (ipa_fn_summary_read): Skip optimize check.
18250 (ipa_fn_summary_write): Likewise.
18251 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18253 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18255 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18256 (check_callers): Give up on caller which is not optimized.
18257 (inline_small_functions): Likewise.
18258 (ipa_inline): Do not give up when not optimizing.
18259 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18260 away unoptimizes cdtors.
18261 (whole_program_function_and_variable_visibility): Do
18262 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18263 * ipa.c (process_references): Do not check optimize.
18264 (symbol_table::remove_unreachable_nodes): Update optimize check.
18265 (set_writeonly_bit): Update optimize check.
18266 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18267 (pass_ipa_single_use::gate): Remove.
18269 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18271 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18272 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18273 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18274 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18275 permute_load, permute_store, adjust_extract, adjust_splat,
18276 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18277 replace_swap_with_copy, dump_swap_insn_table,
18278 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18279 recombine_lvx_pattern, recombine_stvx_pattern,
18280 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18281 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18282 to file rs6000-p8swap.c.
18283 * config/rs6000/rs6000-p8swap.c: New file.
18284 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18285 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18286 and rs6000*-*-* targets.
18288 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18290 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18292 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18294 * lto-wrapper.c (merge_and_complain): Do not merge
18295 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18296 fsigned_zeros, ftrapping_math, fwrapv.
18297 (append_compiler_options): Do not track these options.
18298 (append_linker_options): Likewie
18300 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18302 * cgraphunit.c (cgraph_node::finalize_function): When
18303 !flag_toplevel_reorde set no_reorder flag.
18304 (varpool_node::finalize_decl): Likewise.
18305 (symbol_table::compile): Drop no toplevel reorder path.
18307 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18309 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18310 edges; zero probability is not better than uninitialized.
18312 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18314 * asan.h (asan_sanitize_allocas_p): Declare.
18315 * asan.c (asan_sanitize_allocas_p): New function.
18316 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18317 (handle_builtin_alloca): Likewise.
18318 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18319 if !asan_sanitize_allocas_p.
18320 * params.def (asan-instrument-allocas): Add new option.
18321 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18322 * opts.c (common_handle_option): Disable allocas sanitization for
18325 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18327 * asan.c: Include gimple-fold.h.
18328 (get_last_alloca_addr): New function.
18329 (handle_builtin_stackrestore): Likewise.
18330 (handle_builtin_alloca): Likewise.
18331 (asan_emit_allocas_unpoison): Likewise.
18332 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18333 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18334 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18335 (instrument_builtin_call): Pass gimple iterator to
18336 get_mem_refs_of_builtin_call.
18337 (last_alloca_addr): New global.
18338 * asan.h (asan_emit_allocas_unpoison): Declare.
18339 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18340 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18341 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18342 if function calls alloca.
18343 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18344 * gimple-fold.h (replace_call_with_value): Declare.
18345 * internal-fn.c: Include asan.h.
18346 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18347 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18349 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18351 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18352 (C_SELFTEST_FLAGS): New.
18353 (CPP_SELFTEST_FLAGS): New.
18354 (SELFTEST_DEPS): New, from deps of s-selftest.
18355 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18356 (CPP_SELFTEST_DEPS): New.
18357 (selftest): Add dependency on s-selftest-c++.
18358 (s-selftest): Rename to...
18359 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18360 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18361 than SELFTEST_FLAGS.
18362 (selftest-gdb): Rename to...
18363 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18365 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18366 (selftest-valgrind): Rename to...
18367 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18369 (selftest-valgrind): Reintroduce as an alias for
18370 selftest-c-valgrind.
18371 (s-selftest-c++): New.
18372 (selftest-c++-gdb): New.
18373 (selftest-c++-valgrind): New.
18375 2017-07-06 Olivier Hainque <hainque@adacore.com>
18377 * gcc.c (process_command): When deciding if undefined variables
18378 should be ignored when processing specs, accept "gcc -v" as well.
18380 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18382 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18383 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18385 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18387 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18388 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18389 * config/arm/arm-tables.opt: Regenerate.
18390 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18392 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18394 2017-07-06 Carl Love <cel@us.ibm.com>
18396 * ChangeLog: Clean up from mid air collision
18398 2017-07-06 Carl Love <cel@us.ibm.com>
18400 * config/rs6000/rs6000-c.c: Add support for built-in functions
18401 vector signed int vec_subc (vector signed int, vector signed int);
18402 vector signed __int128 vec_subc (vector signed __int128,
18403 vector signed __int128);
18404 vector unsigned __int128 vec_subc (vector unsigned __int128,
18405 vector unsigned __int128);
18406 vector signed int vec_sube (vector signed int, vector signed int,
18407 vector signed int);
18408 vector unsigned int vec_sube (vector unsigned int,
18409 vector unsigned int,
18410 vector unsigned int);
18411 vector signed __int128 vec_sube (vector signed __int128,
18412 vector signed __int128,
18413 vector signed__int128);
18414 vector unsigned __int128 vec_sube (vector unsigned __int128,
18415 vector unsigned __int128,
18416 vector unsigned __int128);
18417 vector signed int vec_subec (vector signed int, vector signed int,
18418 vector signed int);
18419 vector unsigned int vec_subec (vector unsigned int,
18420 vector unsigned int,
18421 vector unsigned int);
18422 vector signed __int128 vec_subec (vector signed __int128,
18423 vector signed __int128,
18424 vector signed__int128);
18425 vector unsigned __int128 vec_subec (vector unsigned __int128,
18426 vector unsigned __int128,
18427 vector unsigned __int128);
18428 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18429 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18430 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18431 BU_ALTIVEC_OVERLOAD_X definitions.
18432 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18433 * doc/extend.texi: Update the built-in documentation file for the new
18434 built-in functions.
18436 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18439 * diagnostic-show-locus.c (layout::layout): Use start and finish
18440 spelling location for the start and finish of each range.
18441 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18442 Add unused aspect param.
18443 * input.c (expand_location_1): Add "aspect" param, and use it
18444 to access the correct part of the location.
18445 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18447 (expand_location_to_spelling_point): Likewise.
18448 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18449 param, and pass it to expand_location_1.
18451 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18453 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18454 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18455 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18456 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18457 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18458 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18459 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18460 _mm_maskz_getmant_ss): New intrinsics.
18461 (__builtin_ia32_getexpss128_mask): Changed to ...
18462 __builtin_ia32_getexpss128_round ... this.
18463 (__builtin_ia32_getexpsd128_mask): Changed to ...
18464 __builtin_ia32_getexpsd128_round ... this.
18465 * config/i386/i386-builtin-types.def
18466 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18467 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18468 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18469 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18470 __builtin_ia32_getmantss_mask_round): New builtins.
18471 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18472 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18473 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18474 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18475 * config/i386/sse.md
18476 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18477 avx512f_sgetexp<mode><mask_scalar_name>
18478 <round_saeonly_scalar_name> ... this.
18479 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18480 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18481 vgetexp<ssescalarmodesuffix>
18482 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18483 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18484 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18485 avx512f_vgetmant<mode><mask_scalar_name>
18486 <round_saeonly_scalar_name> ... this.
18487 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18488 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18489 vgetmant<ssescalarmodesuffix>
18490 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18491 %0<mask_scalar_operand4>, %1, %2
18492 <round_saeonly_scalar_mask_op4>, %3} ... this.
18493 * config/i386/subst.md (mask_scalar_operand4,
18494 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18495 round_saeonly_scalar_nimm_predicate): New subst attributes.
18497 2017-07-06 Julia Koval <julia.koval@intel.com>
18499 * config/i386/i386.c (ix86_erase_embedded_rounding):
18500 Remove code for old rounding pattern.
18502 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18504 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18506 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18508 * doc/sourcebuild.texi (Test Directives, Variants of
18509 dg-require-support): Add documentation for dg-require-stack-check.
18511 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18513 * config/i386/subst.md (mask_scalar, round_scalar,
18514 round_saeonly_scalar): New meta-templates.
18515 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18516 round_scalar_mask_operand3, round_scalar_mask_op3,
18517 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18518 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18519 round_saeonly_scalar_constraint,
18520 round_saeonly_scalar_prefix): New subst attribute.
18521 * config/i386/sse.md
18522 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18523 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18524 <round_scalar_name> ... this.
18525 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18526 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18527 <round_scalar_name> ... this.
18528 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18529 <sse>_vm<code><mode>3<mask_scalar_name>
18530 <round_saeonly_scalar_name> ... this.
18531 (v<plusminus_mnemonic><ssescalarmodesuffix>
18532 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18533 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18534 v<plusminus_mnemonic><ssescalarmodesuffix>
18535 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18536 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18537 (v<multdiv_mnemonic><ssescalarmodesuffix>
18538 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18539 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18540 v<multdiv_mnemonic><ssescalarmodesuffix>
18541 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18542 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18543 (v<maxmin_float><ssescalarmodesuffix>
18544 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18545 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18546 v<maxmin_float><ssescalarmodesuffix>
18547 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18548 %0<mask_scalar_operand3>, %1, %<iptr>2
18549 <round_saeonly_scalar_mask_op3>} ... this.
18551 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18553 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18554 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18556 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18557 Alan Hayward <alan.hayward@arm.com>
18558 David Sherwood <david.sherwood@arm.com>
18560 * combine.c (simplify_if_then_else): Remove "enum" before
18562 * compare-elim.c (can_eliminate_compare): Likewise.
18563 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18565 (aarch64_lookup_simd_builtin_type): Likewise.
18566 (aarch64_simd_builtin_type): Likewise.
18567 (aarch64_init_simd_builtin_types): Likewise.
18568 (aarch64_simd_expand_args): Likewise.
18569 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18571 (aarch64_reverse_mask): Likewise.
18572 (aarch64_simd_emit_reg_reg_move): Likewise.
18573 (aarch64_gen_adjusted_ldpstp): Likewise.
18574 (aarch64_ccmp_mode_to_code): Likewise.
18575 (aarch64_operands_ok_for_ldpstp): Likewise.
18576 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18577 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18579 (aarch64_min_divisions_for_recip_mul): Likewise.
18580 (aarch64_reassociation_width): Likewise.
18581 (aarch64_get_condition_code_1): Likewise.
18582 (aarch64_simd_emit_reg_reg_move): Likewise.
18583 (aarch64_simd_attr_length_rglist): Likewise.
18584 (aarch64_reverse_mask): Likewise.
18585 (aarch64_operands_ok_for_ldpstp): Likewise.
18586 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18587 (aarch64_gen_adjusted_ldpstp): Likewise.
18588 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18590 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18591 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18592 (arm_lookup_simd_builtin_type): Likewise.
18593 (arm_simd_builtin_type): Likewise.
18594 (arm_init_simd_builtin_types): Likewise.
18595 (arm_expand_builtin_args): Likewise.
18596 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18597 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18598 (ft32_setup_incoming_varargs): Likewise.
18599 (ft32_function_arg): Likewise.
18600 (ft32_function_arg_advance): Likewise.
18601 (ft32_pass_by_reference): Likewise.
18602 (ft32_arg_partial_bytes): Likewise.
18603 (ft32_valid_pointer_mode): Likewise.
18604 (ft32_addr_space_pointer_mode): Likewise.
18605 (ft32_addr_space_legitimate_address_p): Likewise.
18606 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18608 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18609 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18610 (iamcu_alignment): Likewise.
18611 (canonicalize_vector_int_perm): Likewise.
18612 (ix86_noce_conversion_profitable_p): Likewise.
18613 (ix86_mpx_bound_mode): Likewise.
18614 (ix86_operands_ok_for_move_multiple): Likewise.
18615 * config/microblaze/microblaze-protos.h
18616 (microblaze_expand_conditional_branch_reg): Likewise.
18617 * config/microblaze/microblaze.c
18618 (microblaze_expand_conditional_branch_reg): Likewise.
18619 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18621 (rs6000_reassociation_width): Likewise.
18622 (rs6000_invalid_binary_op): Likewise.
18623 (fusion_p9_p): Likewise.
18624 (emit_fusion_p9_load): Likewise.
18625 (emit_fusion_p9_store): Likewise.
18626 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18628 (riscv_hard_regno_mode_ok_p): Likewise.
18629 (riscv_address_insns): Likewise.
18630 (riscv_split_symbol): Likewise.
18631 (riscv_legitimize_move): Likewise.
18632 (riscv_function_value): Likewise.
18633 (riscv_hard_regno_nregs): Likewise.
18634 (riscv_expand_builtin): Likewise.
18635 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18636 (riscv_build_integer): Likewise.
18637 (riscv_split_integer): Likewise.
18638 (riscv_legitimate_constant_p): Likewise.
18639 (riscv_cannot_force_const_mem): Likewise.
18640 (riscv_regno_mode_ok_for_base_p): Likewise.
18641 (riscv_valid_base_register_p): Likewise.
18642 (riscv_valid_offset_p): Likewise.
18643 (riscv_valid_lo_sum_p): Likewise.
18644 (riscv_classify_address): Likewise.
18645 (riscv_legitimate_address_p): Likewise.
18646 (riscv_address_insns): Likewise.
18647 (riscv_load_store_insns): Likewise.
18648 (riscv_force_binary): Likewise.
18649 (riscv_split_symbol): Likewise.
18650 (riscv_force_address): Likewise.
18651 (riscv_legitimize_address): Likewise.
18652 (riscv_move_integer): Likewise.
18653 (riscv_legitimize_const_move): Likewise.
18654 (riscv_legitimize_move): Likewise.
18655 (riscv_address_cost): Likewise.
18656 (riscv_subword): Likewise.
18657 (riscv_output_move): Likewise.
18658 (riscv_canonicalize_int_order_test): Likewise.
18659 (riscv_emit_int_order_test): Likewise.
18660 (riscv_function_arg_boundary): Likewise.
18661 (riscv_pass_mode_in_fpr_p): Likewise.
18662 (riscv_pass_fpr_single): Likewise.
18663 (riscv_pass_fpr_pair): Likewise.
18664 (riscv_get_arg_info): Likewise.
18665 (riscv_function_arg): Likewise.
18666 (riscv_function_arg_advance): Likewise.
18667 (riscv_arg_partial_bytes): Likewise.
18668 (riscv_function_value): Likewise.
18669 (riscv_pass_by_reference): Likewise.
18670 (riscv_setup_incoming_varargs): Likewise.
18671 (riscv_print_operand): Likewise.
18672 (riscv_elf_select_rtx_section): Likewise.
18673 (riscv_save_restore_reg): Likewise.
18674 (riscv_for_each_saved_reg): Likewise.
18675 (riscv_register_move_cost): Likewise.
18676 (riscv_hard_regno_mode_ok_p): Likewise.
18677 (riscv_hard_regno_nregs): Likewise.
18678 (riscv_class_max_nregs): Likewise.
18679 (riscv_memory_move_cost): Likewise.
18680 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18681 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18682 (rl78_addr_space_address_mode): Likewise.
18683 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18685 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18686 (rs6000_reassociation_width): Likewise.
18687 (rs6000_invalid_binary_op): Likewise.
18688 (fusion_p9_p): Likewise.
18689 (emit_fusion_p9_load): Likewise.
18690 (emit_fusion_p9_store): Likewise.
18691 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18692 (ok_for_simple_move_operands): Likewise.
18693 (ok_for_simple_move_strict_operands): Likewise.
18694 (ok_for_simple_arith_logic_operands): Likewise.
18695 (visium_legitimize_reload_address): Likewise.
18696 (visium_select_cc_mode): Likewise.
18697 (output_cbranch): Likewise.
18698 (visium_split_double_move): Likewise.
18699 (visium_expand_copysign): Likewise.
18700 (visium_expand_int_cstore): Likewise.
18701 (visium_expand_fp_cstore): Likewise.
18702 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18703 (visium_function_arg): Likewise.
18704 (visium_function_arg_advance): Likewise.
18705 (visium_libcall_value): Likewise.
18706 (visium_setup_incoming_varargs): Likewise.
18707 (visium_legitimate_constant_p): Likewise.
18708 (visium_legitimate_address_p): Likewise.
18709 (visium_legitimize_address): Likewise.
18710 (visium_secondary_reload): Likewise.
18711 (visium_register_move_cost): Likewise.
18712 (visium_memory_move_cost): Likewise.
18713 (prepare_move_operands): Likewise.
18714 (ok_for_simple_move_operands): Likewise.
18715 (ok_for_simple_move_strict_operands): Likewise.
18716 (ok_for_simple_arith_logic_operands): Likewise.
18717 (visium_function_value_1): Likewise.
18718 (rtx_ok_for_offset_p): Likewise.
18719 (visium_legitimize_reload_address): Likewise.
18720 (visium_split_double_move): Likewise.
18721 (visium_expand_copysign): Likewise.
18722 (visium_expand_int_cstore): Likewise.
18723 (visium_expand_fp_cstore): Likewise.
18724 (visium_split_cstore): Likewise.
18725 (visium_select_cc_mode): Likewise.
18726 (visium_split_cbranch): Likewise.
18727 (output_cbranch): Likewise.
18728 (visium_print_operand_address): Likewise.
18729 * expmed.c (flip_storage_order): Likewise.
18730 * expmed.h (emit_cstore): Likewise.
18731 (flip_storage_order): Likewise.
18732 * genrecog.c (validate_pattern): Likewise.
18733 * hsa-gen.c (gen_hsa_addr): Likewise.
18734 * internal-fn.c (expand_arith_overflow): Likewise.
18735 * ira-color.c (allocno_copy_cost_saving): Likewise.
18736 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18737 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18738 (process_invariant_for_inheritance): Likewise.
18739 * lra-eliminations.c (move_plus_up): Likewise.
18740 * omp-low.c (lower_oacc_reductions): Likewise.
18741 * simplify-rtx.c (simplify_subreg): Likewise.
18742 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18743 (TARGET_CHKP_BOUND_MODE): Likewise..
18744 * targhooks.c (default_chkp_bound_mode): Likewise.
18745 (default_setup_incoming_vararg_bounds): Likewise.
18746 * targhooks.h (default_chkp_bound_mode): Likewise.
18747 (default_setup_incoming_vararg_bounds): Likewise.
18748 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18749 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18750 (have_whole_vector_shift): Likewise.
18751 * tree-vect-stmts.c (vectorizable_load): Likewise.
18752 * doc/tm.texi: Regenerate.
18754 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18756 Graceful degrade if Binutils PR21472 is not available.
18759 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18760 .rodata in flash test fails.
18761 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18762 * confgure: Regenerate.
18763 * config.in: Regenerate.
18764 * config/avr/avr.c (avr_asm_named_section)
18765 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18766 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18767 (avr_asm_init_sections): Same.
18769 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18771 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18772 (fma<VH:mode>4_intrinsic): Likewise.
18773 (*fmsub<VCVTF:mode>4): Likewise.
18774 (*fmsub<VH:mode>4_intrinsic): Likewise.
18776 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18779 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18780 Don't depend on "optimize > 0".
18781 (out_movhi_r_mr, out_movqi_mr_r): Same.
18782 (out_movhi_mr_r, out_movqi_r_mr): Same.
18783 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18784 io_address_operand on "optimize > 0".
18786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18788 * tree-loop-distribution.c: Add general explanantion on the pass.
18789 (generate_loops_for_partition): Mark distributed loop.
18790 (pg_add_dependence_edges): New parameter. Handle alias data
18791 dependence specially and record it in the parameter if asked.
18792 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18793 (init_partition_graph_vertices, add_partition_graph_edge): New.
18794 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18795 (free_partition_graph_vdata, build_partition_graph): New.
18796 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18797 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18798 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18799 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18800 (version_for_distribution_p, finalize_partitions): New.
18801 (distribute_loop): Handle alias data dependence specially. Factor
18802 out loop fusion code as functions and call these functions.
18804 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18806 * tree-loop-distribution.c (classify_partition): New parameter and
18807 better handle reduction statement.
18808 (rdg_build_partitions): Revise comment.
18809 (distribute_loop): Compute statements in all partitions and pass it
18810 to classify_partition.
18812 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18814 * tree-loop-distribution.c (enum partition_type): New.
18815 (struct partition): New field type.
18816 (partition_merge_into): Add parameter. Update partition type.
18817 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18818 (build_rdg_partition_for_vertex): Compute partition type.
18819 (rdg_build_partitions): Dump partition type.
18820 (distribute_loop): Update calls to partition_merge_into.
18822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18824 * tree-loop-distribution.c (struct ddr_hasher): New.
18825 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18827 (classify_partition): Call get_data_dependence.
18828 (pg_add_dependence_edges): Ditto.
18829 (distribute_loop): Release data dependence hash table.
18831 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18833 * tree-loop-distribution.c (ref_base_address): Delete.
18834 (similar_memory_accesses): Rename ...
18835 (share_memory_accesses): ... to this. Check if partitions access
18836 the same memory reference.
18837 (distribute_loop): Call share_memory_accesses.
18839 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18841 * tree-loop-distribution.c (struct partition): New field recording
18842 its data reference.
18843 (partition_alloc, partition_free): Init and release data refs.
18844 (partition_merge_into): Merge data refs.
18845 (build_rdg_partition_for_vertex): Collect data refs for partition.
18846 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18848 (distribute_loop): Remve data refs from vertice data of partition
18851 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18853 * tree-loop-distribution.c (params.h): Include header file.
18854 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18855 (datarefs_vec): New global var.
18856 (create_rdg_vertices): Use datarefs_vec directly.
18857 (free_rdg): Don't free data references.
18858 (build_rdg): Update use. Don't free data references.
18859 (distribute_loop): Compute global variable for data references.
18860 Bail out if there are too many data references.
18862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18864 * tree-loop-distribution.c (loop_nest): New global var.
18865 (build_rdg): Use loop directly, rather than loop nest.
18866 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18868 (distribute_loop): Compute global variable loop nest. Update use.
18870 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18872 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18873 (partition_merge_into): New parameter. Dump reason for fusion.
18874 (distribute_loop): Update use of partition_merge_into.
18876 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18878 * tree-loop-distribution.c (bb_top_order_index): New.
18879 (bb_top_order_index_size, bb_top_order_cmp): New.
18880 (stmts_from_loop): Use topological order.
18881 (pass_loop_distribution::execute): Compute and release topological
18882 order for basic blocks.
18884 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18886 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18889 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18891 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18892 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18893 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18894 * internal-fn.def (LOOP_DIST_ALIAS): New.
18895 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18896 (fold_loop_internal_call): ... this.
18897 (vect_loop_dist_alias_call): New function.
18898 (set_uid_loop_bbs): Call fold_loop_internal_call.
18899 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18902 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18905 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18906 Require dead FLAGS_REG at the beginning of a peephole.
18908 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18911 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18912 arguments in the call to __builtin_ia32_sbb_u32.
18913 (_subborrow_u64): Swap _X and _Y arguments in the call to
18914 __builtin_ia32_sbb_u64.
18916 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18919 * tree-vrp.c (compare_assert_loc): Turn into a function template
18920 with stable template parameter. Only test if a->e is NULL,
18921 !a->e == !b->e has been verified already. Use e == NULL or
18922 e != NULL instead of e or ! e tests. If stable is true, don't use
18923 iterative_hash_expr, on the other side allow a or b or both NULL
18924 and sort the NULLs last.
18925 (process_assert_insertions): Sort using compare_assert_loc<false>
18926 instead of compare_assert_loc, later sort using
18927 compare_assert_loc<true> before calling process_assert_insertions_for
18928 in a loop. Use break instead of continue once seen NULL pointer.
18930 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18932 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18933 Cortex-R7 and Cortex-R8 processors.
18935 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18937 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18938 uninitialized while src is not.
18940 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18942 * common/config/arm/arm-common.c: Adjust include path for
18944 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18945 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18946 (arm-cpu-data.h): Likewise.
18947 (arm-cpu-cdata.h): Likewise.
18948 * config/arm/arm-cpu.h: Delete.
18949 * config/arm/arm-cpu-cdata.h: Delete.
18950 * config/arm/arm-cpu-data.h: Delete.
18952 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18954 * config/arm/arm-cpus.in (cortex-a55): New.
18955 (cortex-a75): Likewise.
18956 (cortex-a75.cortex-a55): Likewise.
18957 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18959 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18960 * config/arm/arm-cpu-cdata.h: Regenerate.
18961 * config/arm/arm-cpu-data.h: Regenerate.
18962 * config/arm/arm-cpu.h: Regenerate.
18963 * config/arm/arm-tables.opt: Regenerate.
18964 * config/arm/arm-tune.md: Regenerate.
18966 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18968 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18970 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18972 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18975 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18977 PR tree-optimization/81292
18978 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18979 full_string_p, also call adjust_related_strinfos if the adjustment
18980 is simple, otherwise invalidate related strinfos.
18982 2017-07-04 Martin Liska <mliska@suse.cz>
18985 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18986 newly created variable as DECL_IGNORED_P.
18988 2017-07-04 Martin Liska <mliska@suse.cz>
18991 * ipa-inline.c (inline_small_functions):
18992 Use xstrdup_for_dump.
18994 2017-07-04 Tom de Vries <tom@codesourcery.com>
18996 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18998 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19001 * config/darwin.c (darwin_function_switched_text_sections):
19004 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19006 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19008 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19010 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19012 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19014 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19015 min_profitable_iters, and th as inclusive lower bounds.
19016 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19017 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19018 for min_profitable_iters and min_profitable_estimate.
19019 (vect_transform_loop): Treat th as an inclusive lower bound.
19020 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19022 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19025 * config/darwin.c (darwin_function_switched_text_sections):
19026 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19027 in two pieces, and suppress the use of buf.
19029 2017-07-03 Nathan Sidwell <nathan@acm.org>
19031 * hash-table.h (hash_table_mod1): Fix indentation.
19033 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19035 PR middle-end/81290
19036 * predict.c (force_edge_cold): Be more careful about propagation
19038 * profile-count.h (profile_probability::guessed,
19039 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19041 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19043 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19045 * doc/invoke.texi (rcpc architecture extension): Document it.
19047 2017-07-03 Richard Biener <rguenther@suse.de>
19049 PR tree-optimization/60510
19050 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19051 the scalar reduction PHI and use it.
19052 (vectorizable_reduction): Properly guard the single_defuse_cycle
19053 path for non-SLP reduction chains where we cannot use it.
19054 Rework reduc_def/index and vector type deduction. Rework
19055 vector operand gathering during reduction op code-gen.
19056 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19057 chains dissolve the chain and leave it to non-SLP reduction
19060 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19062 * tree-data-ref.h (dr_alignment): Declare.
19063 * tree-data-ref.c (dr_alignment): New function.
19064 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19065 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19067 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19069 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19071 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19072 and base_misalignment fields.
19073 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19074 * tree-data-ref.c: Include builtins.h.
19075 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19076 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19077 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19078 * tree-vect-data-refs.c: Include tree-cfg.h.
19079 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19080 fields instead of calculating an alignment here.
19081 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19082 innermost_loop_behavior fields.
19084 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19086 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19088 (DR_STEP_ALIGNMENT): New macro.
19089 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19090 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19091 (create_data_ref): Print it.
19092 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19093 to tell whether the step preserves vector (mis)alignment.
19094 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19095 Move the check for an integer step and generalise to all INTEGER_CST.
19096 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19097 Print the outer step alignment.
19099 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19101 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19102 with offset_alignment.
19103 (DR_ALIGNED_TO): Delete.
19104 (DR_OFFSET_ALIGNMENT): New macro.
19105 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19106 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19107 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19108 (create_data_ref): Likewise.
19109 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19110 (vect_analyze_data_refs): Likewise.
19111 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19112 creating dummy innermost behavior.
19114 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19116 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19117 with a "innermost_loop_behavior *" and refeence tree.
19118 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19119 (create_data_ref): Update call accordingly.
19120 * tree-predcom.c (find_looparound_phi): Likewise.
19122 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19124 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19125 fields with dr_wrt_vec_loop.
19126 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19127 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19128 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19129 (vect_dr_behavior): New function.
19130 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19131 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19132 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19133 track whether the step preserves the misalignment.
19134 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19135 Use vect_dr_behavior.
19136 (vect_setup_realignment): Update call accordingly.
19137 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19138 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19139 call to vect_create_addr_base_for_vector_ref.
19140 (vect_create_cond_for_align_checks): Likewise.
19141 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19142 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19143 (vect_recog_mask_conversion_pattern): Likewise.
19144 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19145 (new_stmt_vec_info): Remove redundant zeroing.
19147 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19149 * common/config/arm/arm-common.c (arm_be8_option): New function.
19150 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19151 (ISA_ARMv6): Add isa_bit_be8.
19152 * config/arm/arm.h (arm_be8_option): Add prototype.
19153 (BE8_SPEC_FUNCTION): New define.
19154 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19155 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19156 (mlittle-endian): Similarly.
19157 (mbe8, mbe32): New options.
19158 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19159 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19161 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19163 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19165 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19167 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19168 (cleanup_tree_cfg_bb): Use it.
19169 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19171 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19176 * loop-doloop.c (add_test): Update profile.
19178 2017-07-03 Martin Liska <mliska@suse.cz>
19181 * sanopt.c (rewrite_usage_of_param): New function.
19182 (sanitize_rewrite_addressable_params): Likewise.
19183 (pass_sanopt::execute): Call rewrite_usage_of_param.
19185 2017-07-03 Richard Biener <rguenther@suse.de>
19187 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19188 back to using VIEW_CONVERT_EXPR.
19190 2017-07-03 Martin Liska <mliska@suse.cz>
19193 * doc/extend.texi: Document when a resolver function is
19194 generated for target_clones.
19196 2017-07-03 Martin Liska <mliska@suse.cz>
19198 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19199 and shadow memory of auto variables which are subject of
19200 use-after-scope sanitization.
19201 (asan_expand_mark_ifn): Add do set only when is_poison.
19203 2016-07-03 Richard Biener <rguenther@suse.de>
19205 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19207 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19208 (vectorizable_reduction): Perform reduction PHI creation when
19209 visiting a reduction PHI and adjust and simplify code generation
19210 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19211 (vect_transform_loop): Visit reduction PHIs.
19212 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19213 defs into the SLP tree.
19214 (vect_build_slp_tree): Reduction defs terminate the recursion.
19215 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19217 (vect_get_vec_defs_for_stmt_copy): Export.
19218 (vect_get_vec_defs): Likewise.
19219 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19221 (vect_get_vec_defs_for_stmt_copy): Declare.
19222 (vect_get_vec_defs): Likewise.
19224 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19226 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19227 parameter with a "loop" parameter and use it instead of the
19228 loop containing DR_STMT. Don't check simple_iv when doing
19229 BB analysis. Describe the two analysis modes in the comment.
19231 2017-07-03 Tom de Vries <tom@codesourcery.com>
19233 PR tree-optimization/69468
19234 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19235 (find_same_succ_bb): Handle ignore_edge_flags.
19237 2017-07-03 Tom de Vries <tom@codesourcery.com>
19239 PR tree-optimization/81192
19240 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19242 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19244 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19246 2017-07-03 Tom de Vries <tom@codesourcery.com>
19248 PR tree-optimization/81192
19249 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19250 BB_SAME_SUCC (bb) == NULL.
19252 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19254 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19257 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19259 * dumpfile.c: Include profile-count.h
19260 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19262 (insert_cond_bb): Update profile.
19263 * tree-cfg.h (insert_cond_bb): Update prototype.
19264 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19265 * tree-dump.c: Do not include tree-cfg.
19267 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19269 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19271 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19273 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19276 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19278 * tree-complex.c (expand_complex_div_wide): update profile.
19280 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19281 Alan Hayward <alan.hayward@arm.com>
19282 David Sherwood <david.sherwood@arm.com>
19284 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19285 (CORETYPES_H): New define.
19286 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19287 (insn-modes-inline.h, s-modes-inline-h): New rules.
19288 (generated_files): Add insn-modes-inline.h.
19289 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19290 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19291 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19292 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19293 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19294 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19295 (build/gencodes.o, build/genconditions.o): Likewise.
19296 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19297 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19298 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19299 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19300 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19301 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19302 * coretypes.h: Include everything up to real.h for generators.
19303 Include insn-modes.h first. Include wide-int-print.h after
19304 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19305 * machmode.h: Don't include insn-modes.h here.
19306 * function-tests.c: Remove includes of signop.h, machmode.h,
19307 double-int.h and wide-int.h.
19309 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19311 * optc-save-gen.awk: Likewise.
19312 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19313 * godump.c: Remove include of wide-int-print.h.
19314 * pretty-print.h: Likewise.
19315 * wide-int-print.cc: Likewise.
19316 * wide-int.cc: Likewise.
19317 * hash-map-tests.c: Remove include of signop.h.
19318 * hash-set-tests.c: Likewise.
19319 * rtl-tests.c: Likewise.
19320 * mkconfig.sh: Remove include of machmode.h.
19321 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19323 (emit_insn_modes_inline_h): ...this new function. Emit the code
19324 into an insn-modes-inline.h header file, adding appropriate
19325 include guards and end comments.
19326 (emit_insn_modes_c_header): Remove include of machmode.h.
19327 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19329 (main): Handle -i flag and call emit_insn_modes_inline_h when
19332 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19334 * tree-ssa-strlen.c (strinfo): Rename the length field to
19335 nonzero_chars. Add a full_string_p field.
19336 (compare_nonzero_chars, zero_length_string_p): New functions.
19337 (get_addr_stridx): Add an offset_out parameter.
19338 Use compare_nonzero_chars.
19339 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19340 (new_strinfo): Update after above changes to strinfo.
19341 (set_endptr_and_length): Set full_string_p.
19342 (get_string_length): Update after above changes to strinfo.
19343 (unshare_strinfo): Update call to new_strinfo.
19344 (maybe_invalidate): Likewise.
19345 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19346 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19347 as a uhwi instead of an shwi. Update after above changes to
19348 strinfo and new_strinfo.
19349 (zero_length_string): Assert that chainsi contains full strings.
19350 Use zero_length_string_p. Update call to new_strinfo.
19351 (adjust_related_strinfos): Update after above changes to strinfo.
19352 Copy full_string_p from origsi.
19353 (adjust_last_stmt): Use zero_length_string_p.
19354 (handle_builtin_strlen): Update after above changes to strinfo and
19355 new_strinfo. Install the lhs as the string length if the previous
19356 entry didn't describe a full string.
19357 (handle_builtin_strchr): Update after above changes to strinfo
19359 (handle_builtin_strcpy): Likewise.
19360 (handle_builtin_strcat): Likewise.
19361 (handle_builtin_malloc): Likewise.
19362 (handle_pointer_plus): Likewise.
19363 (handle_builtin_memcpy): Likewise. Track nonzero characters
19364 that aren't necessarily followed by a nul terminator.
19365 (handle_char_store): Likewise.
19367 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19369 PR tree-optimization/80769
19370 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19371 for malloc and calloc. Document the new invariant that all related
19372 strinfos have delayed lengths or none do.
19373 (verify_related_strinfos): Move earlier in file.
19374 (set_endptr_and_length): New function, split out from...
19375 (get_string_length): ...here. Also set the lengths of related
19377 (zero_length_string): Assert that chainsi has known (rather than
19379 (adjust_related_strinfos): Likewise.
19381 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19383 PR tree-optimization/81136
19384 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19385 assert that two references with the same misalignment have the same
19386 compile-time misalignment if those compile-time misalignments
19389 2017-07-01 Andi Kleen <ak@linux.intel.com>
19391 * print-tree.c (print_node): Print all attributes.
19393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19395 * cfg.c (scale_bbs_frequencies): New function.
19396 * cfg.h (scale_bbs_frequencies): Declare it.
19397 * cfgloopanal.c (single_likely_exit): Cleanup.
19398 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19400 (scale_loop_profile): Likewise.
19401 (loop_version): Likewise.
19402 (create_empty_loop_on_edge): Update.
19403 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19404 scale_loop_frequencies, scale_loop_profile, loopify,
19405 loop_version): Update prototypes.
19406 * modulo-sched.c (sms_schedule): Update.
19407 * predict.c (unlikely_executed_edge_p): Also check probability.
19408 (probably_never_executed_edge_p): Fix typo.
19409 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19410 * tree-parloops.c (gen_parallel_loop): Update.
19411 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19412 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19413 * tree-ssa-loop-split.c (split_loop): Update.
19414 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19415 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19416 (vect_loop_versioning): Update.
19417 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19419 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19421 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19423 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19425 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19426 to keep profile consistent.
19428 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19430 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19431 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19432 * profile-count.h (max_safe_multiplier): Make unsigned.
19433 (profile_count::guessed_zero): New.
19435 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19437 * bb-reorder.c (fix_up_crossing_landing_pad,
19438 fix_crossing_conditional_branches): Use make_single_succ_edge
19439 to keep profile consistent.
19441 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19443 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19446 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19449 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19450 the right scopes, make sure cond_jump isn't preserved between multiple
19451 iterations. Search for fallthru edge whenever there are 3+ edges and
19452 use find_fallthru_edge for it.
19454 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19456 Patch by Alexander Monakov <amonakov@ispras.ru>
19457 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19458 probabilities consistently.
19460 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19462 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19463 emit_cmp_and_jump_insns.
19465 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19468 * tree-inline.c (expand_call_inline): Combine profile statuses.
19470 2017-06-30 Andrew Pinski <apinski@cavium.com>
19472 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19473 fold_stmt returned true.
19475 2017-06-30 Nathan Sidwell <nathan@acm.org>
19477 * ggc.h (empty_string): Delete.
19478 * cfgexpand.c (expand_asm_stmt): Use plain "".
19479 * optabs.c (expand_asm_memory_barrier): Likewise.
19480 * stringpool.c (empty_string): Delete.
19481 (digit_vector, digit_string): Delete.
19482 (ggc_alloc_string): Use plain "", don't optimize single digit
19483 strings. Use ggc_alloc_atomic.
19485 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19487 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19488 comparison set and one other set, use the cost of the non-comparison
19491 2017-06-30 Nathan Sidwell <nathan@acm.org>
19493 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19496 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19498 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19499 loops. Remove now unneeded calls to gimple_switch_set_label() that
19500 just set removed labels to NULL_TREE.
19502 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19504 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19505 (set_range_info): ...here. Only call set_range_info_raw if domain
19507 (set_nonzero_bits): Call set_range_info_raw.
19508 * tree-ssanames.h (set_range_info_raw): New.
19510 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19513 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19514 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19515 of nonimmediate_operand and <store_mask_constraint> instead of m
19516 for the input operand. For V8FI iterator, always split if input
19517 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19518 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19519 <store_mask_predicate> instead of register_operand and
19520 <store_mask_constraint> instead of v for the input operand. Make
19521 sure both operands aren't MEMs for if not <mask_applied>.
19523 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19525 * lto-wrapper.c (copy_file) Close both file descriptors before
19528 2017-06-30 Martin Liska <mliska@suse.cz>
19531 * multiple_target.c (create_dispatcher_calls): Make ifunc
19532 also for function that don't have calls or are not referenced.
19534 2017-06-30 Richard Biener <rguenther@suse.de>
19536 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19537 analyze the first scalar stmt. Move vector type computation
19538 for the BB case here from ...
19539 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19540 live operation processing in the SLP case properly.
19542 2017-06-30 Richard Biener <rguenther@suse.de>
19544 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19546 2017-06-30 Martin Liska <mliska@suse.cz>
19549 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19550 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19552 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19554 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19557 2017-06-30 Richard Biener <rguenther@suse.de>
19559 PR tree-optimization/81249
19560 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19561 condition reduction result to original scalar type.
19563 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19565 * profile-count.h (enum profile_quality): Fix typos and whitespace
19568 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19570 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19571 type for branch probabilities.
19573 2017-06-29 Julian Brown <julian@codesourcery.com>
19574 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19576 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19577 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19578 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19579 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19581 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19583 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19584 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19585 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19586 CC usage from generic code to here.
19587 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19588 CC usage into the target macros.
19590 2017-06-29 Maya Rashish <coypu@sdf.org>
19592 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19595 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19597 * arm/arm-builtins.c: Include profile-count.h
19598 * except.c (sjlj_emit_function_enter): Use
19599 profile_probability::unlikely.
19601 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19603 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19604 and tocrel_offset be pointer args rather than implicitly using
19606 (legitimate_constant_pool_address_p, rs6000_emit_move,
19607 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19608 tocrel_offset and use in toc_relative_expr_p call.
19609 (print_operand, print_operand_address): Use static tocrel_base_oac
19610 and tocrel_offset_oac.
19611 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19614 2017-06-29 Maya Rashish <coypu@sdf.org>
19616 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19618 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19620 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19621 objects, take into account only the alignment of 'op0' and 'mode1' if
19624 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19626 * ccmp.c (ccmp_tree_comparison_p): New function.
19627 (ccmp_candidate_p): Update to use above function.
19628 (get_compare_parts): New function.
19629 (expand_ccmp_next): Update to use new functions.
19630 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19632 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19633 take mode as argument.
19634 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19635 * expr.c (expand_expr_real_1): Pass mode as argument.
19637 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19639 * combine.c (combine_instructions): Print insns to dump_file, together
19642 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19644 * asan.c (asan_emit_stack_protection): Update.
19645 (create_cond_insert_point): Update.
19646 * auto-profile.c (afdo_propagate_circuit): Update.
19647 * basic-block.h (struct edge_def): Turn probability to
19648 profile_probability.
19649 (EDGE_FREQUENCY): Update.
19650 * bb-reorder.c (find_traces_1_round): Update.
19651 (better_edge_p): Update.
19652 (sanitize_hot_paths): Update.
19653 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19654 (make_single_succ_edge): Update.
19655 (check_bb_profile): Update.
19656 (dump_edge_info): Update.
19657 (update_bb_profile_for_threading): Update.
19658 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19659 probabilitycount to 0.
19660 * cfgbuild.c (compute_outgoing_frequencies): Update.
19661 * cfgcleanup.c (try_forward_edges): Update.
19662 (outgoing_edges_match): Update.
19663 (try_crossjump_to_edge): Update.
19664 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19665 (expand_gimple_tailcall): Update.
19666 (construct_init_block): Use make_single_succ_edge.
19667 (construct_exit_block): Use make_single_succ_edge.
19668 * cfghooks.c (verify_flow_info): Update.
19669 (redirect_edge_succ_nodup): Update.
19670 (split_edge): Update.
19671 (account_profile_record): Update.
19672 * cfgloopanal.c (single_likely_exit): Update.
19673 * cfgloopmanip.c (scale_loop_profile): Update.
19674 (set_zero_probability): Remove.
19675 (duplicate_loop_to_header_edge): Update.
19676 * cfgloopmanip.h (loop_version): Update prototype.
19677 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19678 (force_nonfallthru_and_redirect): Update.
19679 (update_br_prob_note): Update.
19680 (rtl_verify_edges): Update.
19681 (purge_dead_edges): Update.
19682 (rtl_lv_add_condition_to_bb): Update.
19683 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19684 * cgraphunit.c (init_lowered_empty_function): Update.
19685 (cgraph_node::expand_thunk): Update.
19686 * cilk-common.c: Include profile-count.h
19687 * dojump.c (inv): Remove.
19688 (jumpifnot): Update.
19689 (jumpifnot_1): Update.
19690 (do_jump_1): Update.
19692 (do_jump_by_parts_greater_rtx): Update.
19693 (do_compare_rtx_and_jump): Update.
19694 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19695 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19696 * dwarf2cfi.c: Include profile-count.h
19697 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19698 (sjlj_emit_dispatch_table): Likewise.
19699 * explow.c: Include profile-count.h
19700 * expmed.c (emit_store_flag_force): Update.
19701 (do_cmp_and_jump): Update.
19702 * expr.c (compare_by_pieces_d::generate): Update.
19703 (compare_by_pieces_d::finish_mode): Update.
19704 (emit_block_move_via_loop): Update.
19705 (store_expr_with_bounds): Update.
19706 (store_constructor): Update.
19707 (expand_expr_real_2): Update.
19708 (expand_expr_real_1): Update.
19709 * expr.h (try_casesi, try_tablejump): Update prototypes.
19710 * gimple-pretty-print.c (dump_probability): Update.
19711 (dump_profile): New.
19712 (dump_gimple_label): Update.
19713 (dump_gimple_bb_header): Update.
19714 * graph.c (draw_cfg_node_succ_edges): Update.
19715 * hsa-gen.c (convert_switch_statements): Update.
19716 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19717 (find_if_case_1): Update.
19718 (find_if_case_2): Update.
19719 * internal-fn.c (expand_arith_overflow_result_store): Update.
19720 (expand_addsub_overflow): Update.
19721 (expand_neg_overflow): Update.
19722 (expand_mul_overflow): Update.
19723 (expand_vector_ubsan_overflow): Update.
19724 * ipa-cp.c (good_cloning_opportunity_p): Update.
19725 * ipa-split.c (split_function): Use make_single_succ_edge.
19726 * ipa-utils.c (ipa_merge_profiles): Update.
19727 * loop-doloop.c (add_test): Update.
19728 (doloop_modify): Update.
19729 * loop-unroll.c (compare_and_jump_seq): Update.
19730 (unroll_loop_runtime_iterations): Update.
19731 * lra-constraints.c (lra_inheritance): Update.
19732 * lto-streamer-in.c (input_cfg): Update.
19733 * lto-streamer-out.c (output_cfg): Update.
19734 * mcf.c (adjust_cfg_counts): Update.
19735 * modulo-sched.c (sms_schedule): Update.
19736 * omp-expand.c (expand_omp_for_init_counts): Update.
19737 (extract_omp_for_update_vars): Update.
19738 (expand_omp_ordered_sink): Update.
19739 (expand_omp_for_ordered_loops): Update.
19740 (expand_omp_for_generic): Update.
19741 (expand_omp_for_static_nochunk): Update.
19742 (expand_omp_for_static_chunk): Update.
19743 (expand_cilk_for): Update.
19744 (expand_omp_simd): Update.
19745 (expand_omp_taskloop_for_outer): Update.
19746 (expand_omp_taskloop_for_inner): Update.
19747 * omp-simd-clone.c (simd_clone_adjust): Update.
19748 * optabs.c (expand_doubleword_shift): Update.
19749 (expand_abs): Update.
19750 (emit_cmp_and_jump_insn_1): Update.
19751 (expand_compare_and_swap_loop): Update.
19752 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19753 * predict.c (predictable_edge_p): Update.
19754 (edge_probability_reliable_p): Update.
19755 (set_even_probabilities): Update.
19756 (combine_predictions_for_insn): Update.
19757 (combine_predictions_for_bb): Update.
19758 (propagate_freq): Update.
19759 (estimate_bb_frequencies): Update.
19760 (force_edge_cold): Update.
19761 * profile-count.c (profile_count::dump): Add missing space into dump.
19762 (profile_count::debug): Add newline.
19763 (profile_count::differs_from_p): Explicitly convert to unsigned.
19764 (profile_count::stream_in): Update.
19765 (profile_probability::dump): New member function.
19766 (profile_probability::debug): New member function.
19767 (profile_probability::differs_from_p): New member function.
19768 (profile_probability::differs_lot_from_p): New member function.
19769 (profile_probability::stream_in): New member function.
19770 (profile_probability::stream_out): New member function.
19771 * profile-count.h (profile_count_quality): Rename to ...
19772 (profile_quality): ... this one.
19773 (profile_probability): New.
19774 (profile_count): Update.
19775 * profile.c (compute_branch_probabilities): Update.
19776 * recog.c (peep2_attempt): Update.
19777 * sched-ebb.c (schedule_ebbs): Update.
19778 * sched-rgn.c (find_single_block_region): Update.
19779 (compute_dom_prob_ps): Update.
19780 (schedule_region): Update.
19781 * sel-sched-ir.c (compute_succs_info): Update.
19782 * stmt.c (struct case_node): Update.
19783 (do_jump_if_equal): Update.
19784 (get_outgoing_edge_probs): Update.
19785 (conditional_probability): Update.
19786 (emit_case_dispatch_table): Update.
19787 (expand_case): Update.
19788 (expand_sjlj_dispatch_table): Update.
19789 (emit_case_nodes): Update.
19790 * targhooks.c: Update.
19791 * tracer.c (better_p): Update.
19792 (find_best_successor): Update.
19793 * trans-mem.c (expand_transaction): Update.
19794 * tree-call-cdce.c: Update.
19795 * tree-cfg.c (gimple_split_edge): Upate.
19796 (move_sese_region_to_fn): Upate.
19797 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19798 * tree-eh.c (lower_resx): Upate.
19799 (cleanup_empty_eh_move_lp): Upate.
19800 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19801 * tree-inline.c (copy_edges_for_bb): Update.
19802 (copy_cfg_body): Update.
19803 * tree-parloops.c (gen_parallel_loop): Update.
19804 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19805 (gimple_gen_time_profiler): Update.
19806 * tree-ssa-dce.c (remove_dead_stmt): Update.
19807 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19808 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19809 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19810 (unloop_loops): Update.
19811 (try_peel_loop): Update.
19812 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19813 * tree-ssa-loop-split.c (connect_loops): Update.
19814 (split_loop): Update.
19815 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19816 (hoist_guard): Update.
19817 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19818 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19819 (value_replacement): Update.
19820 * tree-ssa-reassoc.c (branch_fixup): Update.
19821 * tree-ssa-tail-merge.c (replace_block_by): Update.
19822 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19823 (create_edge_and_update_destination_phis): Update.
19824 (compute_path_counts): Update.
19825 (recompute_probabilities): Update.
19826 (update_joiner_offpath_counts): Update.
19827 (freqs_to_counts_path): Update.
19828 (duplicate_thread_path): Update.
19829 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19830 (struct switch_conv_info): Update.
19831 (gen_inbound_check): Update.
19832 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19833 (vect_do_peeling): Update.
19834 (vect_loop_versioning): Update.
19835 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19836 (optimize_mask_stores): Update.
19837 * ubsan.c (ubsan_expand_null_ifn): Update.
19838 * value-prof.c (gimple_divmod_fixed_value): Update.
19839 (gimple_divmod_fixed_value_transform): Update.
19840 (gimple_mod_pow2): Update.
19841 (gimple_mod_pow2_value_transform): Update.
19842 (gimple_mod_subtract): Update.
19843 (gimple_mod_subtract_transform): Update.
19844 (gimple_ic): Update.
19845 (gimple_stringop_fixed_value): Update.
19846 (gimple_stringops_transform): Update.
19847 * value-prof.h: Update.
19849 2017-06-29 Carl Love <cel@us.ibm.com>
19851 * config/rs6000/rs6000-c.c: Add support for built-in functions
19852 vector signed int vec_signed (vector float);
19853 vector signed long long vec_signed (vector double);
19854 vector signed int vec_signed2 (vector double, vector double);
19855 vector signed int vec_signede (vector double);
19856 vector signed int vec_signedo (vector double);
19857 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19858 instruction generator.
19859 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19860 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19861 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19863 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19864 vunsignede_v2df): Add define_expands.
19865 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19866 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19867 VEC_UNSIGNEDO): Add definitions.
19868 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19869 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19870 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19871 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19872 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19873 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19874 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19875 vec_unsignede, vec_unsignedo): Add builtin defines.
19876 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19878 * doc/extend.texi: Update the built-in documentation file for the
19879 new built-in functions.
19881 2017-06-29 Richard Biener <rguenther@suse.de>
19883 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19884 reduction chains to LOOP_VINFO_REDUCTIONS.
19885 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19886 SLP reductions after processing reduction chains.
19888 2017-06-29 Nathan Sidwell <nathan@acm.org>
19890 * builtins.c (fold_builtin_FUNCTION): Use
19891 lang_hooks.decl_printable_name.
19893 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19895 PR middle-end/81194
19896 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19897 with only one label.
19898 * stmt.c (expand_case): Assert NCASES is greater than one.
19900 2017-06-29 Richard Biener <rguenther@suse.de>
19902 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19904 (group_case_labels): Likewise.
19905 (find_taken_edge): Push sanity checking on val to workers...
19906 (find_taken_edge_cond_expr): ... here
19907 (find_taken_edge_switch_expr): ... and here, handle cases
19908 with just a default label.
19909 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19910 (group_case_labels): Likewise.
19911 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19912 group_case_labels does anything cleanup the CFG again.
19914 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19916 PR tree-optimization/81196
19917 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19918 exit condition comparing two IVs.
19920 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19922 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19923 profile to the dummy entry at the end of the list of architectures.
19924 * config/arm/arm-cpu-cdata.h: Regenerated.
19926 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19927 Michael Collison <michael.collison@arm.com>
19930 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19932 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19933 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19934 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19935 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19936 of shift when the shift amount is masked with constant equal to
19937 the size of the mode.
19938 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19941 2017-06-29 Martin Liska <mliska@suse.cz>
19943 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19945 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19948 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19951 (edge_growth_cache_entry::edge_growth_cache_entry): New
19953 (reset_edge_growth_cache): Update to use constructor.
19955 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19957 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19958 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19959 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19961 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19963 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19964 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19966 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19968 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19969 (*-linux-uclibc*): Add t-uclibc tmake_file.
19970 * config/t-musl: New.
19971 * config/t-uclibc: New.
19973 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19975 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19977 (gen_comm_data): Emit architectural setting of arch_prof.
19978 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19980 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19981 (armv8-m.base, armv8-m.main): Likewise.
19982 * arm-protos.h (arm_build_target): Add profile field.
19983 (arch_option): Likewise.
19984 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19986 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19987 arm_active_target.profile.
19989 2017-06-28 Richard Biener <rguenther@suse.de>
19991 PR middle-end/81227
19992 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19993 TYPE_OVERFLOW_WRAPS.
19994 * match.pd (negate_expr_p): Likewise.
19995 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19996 fold_build2, not fold_binary.
19998 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20000 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20001 Convert memory address to Pmode.
20002 (aarch64_print_operand): Assert MEM operands are always Pmode.
20004 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20007 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20008 Remove redundant if.
20009 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20010 * config/arm/aarch-common-protos.h
20011 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20012 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20014 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20017 * multiple_target.c (create_dispatcher_calls): Set the default
20018 clone to be static, not public.
20020 2017-06-28 Richard Biener <rguenther@suse.de>
20022 * tree-vect-loop.c (vectorizable_reduction): Move special
20023 cond reduction IV var creation ...
20024 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20025 parameter. Use STMT_VINFO_VECTYPE.
20026 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20029 2017-06-28 Martin Liska <mliska@suse.cz>
20032 * ipa-visibility.c (non_local_p): Handle visibility.
20034 2017-06-28 Martin Liska <mliska@suse.cz>
20037 * common.opt: Add IntegerRange to various options.
20038 * opt-functions.awk (integer_range_info): New function.
20039 * optc-gen.awk: Add integer_range_info to cl_options struct.
20040 * opts-common.c (decode_cmdline_option): Handle
20041 CL_ERR_INT_RANGE_ARG.
20042 (cmdline_handle_error): Likewise.
20043 * opts.c (print_filtered_help): Show valid interval in
20044 when --help is provided.
20045 * opts.h (struct cl_option): Add range_min and range_max fields.
20046 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20048 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20050 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20051 (x * C EQ/NE y * C): New transformation.
20053 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20055 * genmultilib (combination_space): Accept '+' in option names.
20057 2017-06-28 Martin Liska <mliska@suse.cz>
20060 * asan.c (instrument_derefs): Bail out inner references
20061 that are hard register variables.
20063 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20066 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20067 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20069 2017-06-28 Richard Biener <rguenther@suse.de>
20071 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20072 (vect_get_slp_defs): Adjust.
20073 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20075 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20077 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20078 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20079 (vectorizable_reduction): Adjust.
20080 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20082 (vect_get_slp_defs): Likewise.
20083 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20084 (vectorizable_bswap): Adjust.
20085 (vectorizable_call): Likewise.
20086 (vectorizable_conversion): Likewise.
20087 (vectorizable_assignment): Likewise.
20088 (vectorizable_shift): Likewise.
20089 (vectorizable_operation): Likewise.
20090 (vectorizable_store): Likewise.
20091 (vectorizable_condition): Likewise.
20092 (vectorizable_comparison): Likewise.
20094 2017-06-28 Michael Collison <michael.collison@arm.com>
20097 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20099 (arm_gen_movmemqi): Removed unused variable 'i'.
20100 Convert 'for' loop into 'while' loop.
20101 (arm_expand_prologue): Remove last unnecessary set of insn.
20102 (thumb_pop): Remove unused variable 'pushed_words'.
20103 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20105 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20107 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20108 * config/s390/s390-protos.h: Add prototype of
20109 s390_rel_address_ok_p.
20110 * config/s390/s390.c (s390_got_symbol): New function.
20111 (s390_rel_address_ok_p): New function.
20112 (legitimize_pic_address): Use s390_rel_address_ok_p.
20113 (s390_load_got): Use s390_got_symbol.
20114 (s390_option_override): Issue error if
20115 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20116 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20118 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20120 2017-06-27 Andrew Pinski <apinski@cavium.com>
20122 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20123 (X * copysign (1.0, X)): New pattern.
20124 (X * copysign (1.0, -X)): New pattern.
20125 (copysign (-1.0, CST)): New pattern.
20127 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20129 * genmultilib (combination_space): Remove variable.
20130 Validate reuse rules against regular expression for any sequence
20131 of multilib options in any order.
20133 2017-06-27 Michael Collison <michael.collison@arm.com>
20135 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20136 call aarch64_split_simd_combine.
20137 * (aarch64_combine_internal<mode>): Delete pattern.
20138 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20139 Allow register and subreg operands.
20141 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20143 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20144 specific need, just fallback on defaults.
20145 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20147 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20148 Olivier Hainque <hainque@adacore.com>
20150 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20152 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20153 targets. Pick a default if no particular attempt applied.
20154 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20157 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20159 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20160 (x86_64-wrs-vxworks7): Likewise.
20162 2017-06-27 Marek Polacek <polacek@redhat.com>
20165 * ubsan.c (instrument_null): Check get_base_address's result for null.
20167 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20169 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20171 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20173 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20174 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20175 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20176 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20177 New function types.
20178 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20179 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20180 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20181 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20182 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20183 BUILT_IN_FEUPDATEENV): New builtins.
20184 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20185 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20186 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20187 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20189 (builtin_structptr_types): Adjust size.
20190 * tree.c (builtin_structptr_types): Add four entries.
20192 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20193 Olivier Hainque <hainque@adacore.com>
20195 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20196 (TLS_SYM): New local macro, forcing reference to __tls__ on
20197 link command lines for VxWorks 7 RTPs, triggering initialization
20199 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20200 OS features TLS support, true for RTPs on VxWorks 7.
20201 * config/vxworks.c (vxworks_override_options): Setup emutls
20204 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20206 * predict.c (test_prediction_value_range): Use -1U instead of -1
20207 to avoid narrowing conversion warning.
20208 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20209 to avoid narrowing conversion warning.
20210 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20212 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20214 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20216 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20217 64bit configurations.
20218 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20219 (SIZE_TYPE): Likewise.
20220 * config/vxworks.c (vxworks_emutls_var_fields): Use
20221 long_unsigned_type_node instead of unsigned_type_node as the offset
20222 field type, which is "pointer" mode in emutls.c.
20224 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20227 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20229 PR middle-end/81207
20230 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20231 gimple_vuse copying separately from gimple_vdef copying.
20233 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20235 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20238 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20239 Olivier Hainque <hainque@adacore.com>
20241 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20242 for all vxworks7 targets.
20243 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20244 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20245 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20246 variations for VX6/VX7 and 32/64bits later on in ...
20247 (VXWORKS_LIB_SPEC): Leverage new macros.
20248 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20249 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20251 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20253 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20254 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20256 2017-06-26 Carl Love <cel@us.ibm.com>
20258 * config/rs6000/rs6000-c.c: Add support for built-in functions
20259 vector bool char vec_reve (vector bool char);
20260 vector signed char vec_reve (vector signed char);
20261 vector unsigned char vec_reve (vector unsigned char);
20262 vector bool int vec_reve (vector bool int);
20263 vector signed int vec_reve (vector signed int);
20264 vector unsigned int vec_reve (vector unsigned int);
20265 vector bool long long vec_reve (vector bool long long);
20266 vector signed long long vec_reve (vector signed long long);
20267 vector unsigned long long vec_reve (vector unsigned long long);
20268 vector bool short vec_reve (vector bool short);
20269 vector signed short vec_reve (vector signed short);
20270 vector double vec_reve (vector double);
20271 vector float vec_reve (vector float);
20272 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20273 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20274 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20275 (altivec_vreve): New pattern.
20276 * config/rs6000/altivec.h (vec_reve): New define.
20277 * doc/extend.texi (vec_rev): Update the built-in documentation file
20278 for the new built-in functions.
20280 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20282 PR tree-optimization/71815
20283 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20285 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20287 (slsr_process_phi): Likewise.
20288 (replace_uncond_cands_and_profitable_phis): Don't replace a
20289 multiply candidate with a stride of 1 (copy or cast).
20290 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20292 (lowest_cost_path): Likewise.
20293 (total_savings): Likewise.
20295 2017-06-26 Richard Biener <rguenther@suse.de>
20298 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20299 Use def_builtin_pure for all gather builtins.
20301 2017-06-26 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/81203
20304 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20305 non-dominating BBs.
20307 2017-06-26 Marek Polacek <polacek@redhat.com>
20310 * doc/invoke.texi: Document -Wmultistatement-macros.
20312 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20314 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20315 arm_neon_ok_no_float_abi effective target.
20317 2017-06-26 Richard Biener <rguenther@suse.de>
20319 PR tree-optimization/80928
20320 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20321 (copy_bbs): Set BB_DUPLICATED flag early.
20322 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20324 (execute_on_shrinking_pred): Likewise.
20325 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20326 BB_DUPLICATED blocks.
20327 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20328 iterate over all PHIs considering removal of *gsi.
20330 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20332 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20335 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20337 * config/rs6000/rs6000-string.c: (expand_block_clear,
20338 do_load_for_compare, select_block_compare_mode,
20339 compute_current_alignment, expand_block_compare,
20340 expand_strncmp_align_check, expand_strn_compare,
20341 expand_block_move, rs6000_output_load_multiple)
20342 Move functions related to string/block move/compare
20343 to a separate file.
20344 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20345 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20346 for this function which is now used in two files.
20347 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20348 * config.gcc: Add rs6000-string.o to extra_objs for
20349 targets powerpc*-*-* and rs6000*-*-*.
20351 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20354 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20355 32-bit, since indexed is not valid for DImode.
20356 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20357 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20358 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20359 (define_peephole2 for Altivec d-form store): Likewise.
20362 * multiple_target.c (create_dispatcher_calls): Only create the
20363 dispatcher call if the function is the default clone of a
20364 versioned function.
20366 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20368 PR middle-end/80902
20369 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20370 a call, force the call to not be a tail call.
20372 2017-06-23 Jeff Law <law@redhat.com>
20374 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20377 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20379 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20380 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20381 handling for early expansion of vector shifts (sl,sr,sra,rl).
20382 (builtin_function_type): Add vector shift right instructions
20383 to the unsigned argument list.
20385 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20387 rtl-optimizatoin/79286
20388 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20389 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20390 trap. PIC register plus a const unspec without offset can never trap.
20392 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20394 * tree.h (builtin_structptr_type): New type.
20395 (builtin_structptr_types): Declare new array.
20396 * tree.c (builtin_structptr_types): New array.
20397 (free_lang_data, build_common_tree_nodes): Use it.
20399 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20402 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20405 2017-06-22 Matt Turner <mattst88@gmail.com>
20407 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20408 Lake models to skylake case. Assume skylake for unknown
20409 models with clflushopt.
20411 2017-06-22 Jeff Law <law@redhat.com>
20413 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20414 frame sizes that do not satisfy aarch64_uimm12_shift.
20416 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20418 * profile-count.h (apply_probability,
20419 apply_scale, probability_in): Fix checks for zero.
20421 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20423 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20424 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20426 2016-06-22 Richard Biener <rguenther@suse.de>
20428 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20429 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20430 REDUC_MAX_EXPR support.
20431 (vectorizable_reduction): Likewise.
20432 (vect_create_epilog_for_reduction): Likewise.
20434 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20436 * match.pd (A / (1 << B) -> A >> B): New.
20437 * generic-match-head.c: Include optabs-tree.h.
20438 * gimple-match-head.c: Likewise.
20439 * optabs-tree.h (target_supports_op_p): New.
20440 * optabs-tree.c (target_supports_op_p): New.
20442 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20444 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20445 $gcc_cv_ld --help output.
20446 (gcc_cv_ld_demangle): Likewise.
20447 (gcc_cv_ld_eh_frame_hdr): Likewise.
20448 (gcc_cv_ld_pie): Likewise.
20449 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20450 (gcc_cv_ld_buildid): Likewise.
20451 (gcc_cv_ld_sysroot): Likewise.
20452 (ld_bndplt_support): Likewise.
20453 (ld_pushpopstate_support): Likewise.
20454 * configure: Regenerate.
20455 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20457 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20460 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20461 operands indexes to avoid gap between operands and match_dups.
20463 2017-06-21 Andrew Pinski <apinski@cavium.com>
20465 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20466 Increment Arith_shift and Arith_shift_reg by 1.
20467 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20469 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20470 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20471 (aarch64_strip_extend): Add new argument and test for it.
20472 (aarch64_cheap_mult_shift_p): New function.
20473 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20474 add a cost if it is true.
20475 Update calls to aarch64_strip_extend.
20476 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20478 2017-06-21 Andrew Pinski <apinski@cavium.com>
20480 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20482 (thunderxt88): Likewise.
20483 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20484 (thunderx_prefetch_tune): New variable.
20485 (thunderx2t99_prefetch_tune): Update for the correct values.
20486 (thunderxt88_tunings): New variable.
20487 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20488 generic_prefetch_tune.
20489 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20491 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20493 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20494 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20495 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20496 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20497 (aarch64_atomic_cas<mode>, GPI): Likewise.
20499 2017-06-21 Martin Liska <mliska@suse.cz>
20501 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20502 statements on cold and hot labels.
20503 * predict.c (tree_estimate_probability_bb): Remove the
20504 prediction from this place.
20506 2017-06-21 Martin Liska <mliska@suse.cz>
20508 PR tree-optimization/79489
20509 * gimplify.c (maybe_add_early_return_predict_stmt): New
20511 (gimplify_return_expr): Call the function.
20512 * predict.c (tree_estimate_probability_bb): Remove handling
20514 * predict.def: Update comment about early return predictor.
20515 * gimple-predict.h (is_gimple_predict): New function.
20516 * predict.def: Change default value of early return to 66.
20517 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20519 * passes.def: Put pass_strip_predict_hints to the beginning of
20522 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20524 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20525 FUNCTION_DECL declarations.
20526 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20528 (dwaf2out_decl): Likewise.
20529 * godump.c (go_early_global_decl): Skip call to the real debug hook
20530 for FUNCTION_DECL declarations.
20531 * passes.c (rest_of_decl_compilation): Skip call to the
20532 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20533 -fdump-go-spec is passed.
20535 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20537 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20538 Reorder for compactness.
20539 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20540 (def_builtin_pure, def_builtin_pure2): New functions.
20541 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20543 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20545 * match.pd (nop_convert): New predicate.
20546 ((A +- CST1) +- CST2): Allow some NOP conversions.
20548 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20551 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20552 with ctors/dtors if GOVD_SHARED is set.
20554 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20556 * config/aarch64/aarch64.md (movti_aarch64):
20557 Emit mov rather than orr.
20558 (movtf_aarch64): Likewise.
20559 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20560 Emit mov rather than orr.
20562 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20564 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20565 Swap alternatives, make integer dup more expensive.
20567 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20569 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20570 Return true for non-tls symbols.
20572 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20574 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20575 (cortex-a75): Likewise.
20576 (cortex-a75.cortex-a55): Likewise.
20577 * config/aarch64/aarch64-tune.md: Regenerate.
20578 * doc/invoke.texi (-mtune): Document new values for -mtune.
20580 2017-06-21 Tom de Vries <tom@codesourcery.com>
20582 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20583 stack_size feature.
20584 (Effective-Target Keywords, Other attributes): Suggest using
20585 dg-add-options stack_size feature to get stack limit in stack_size
20586 effective target documentation.
20588 2017-06-21 Julian Brown <julian@codesourcery.com>
20589 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20591 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20592 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20593 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20595 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20596 attribute type list for neon_multiply.
20597 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20598 attribute type list for neon_multiply.
20599 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20600 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20601 attribute type list for neon_multiply.
20602 * config/arm/types.md (crypto_pmull): Add.
20603 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20604 attribute type list.
20606 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20608 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20611 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20613 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20614 to make sure not to dereference a NULL cost_classes_ptr pointer.
20616 2017-06-20 Carl Love <cel@us.ibm.com>
20618 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20619 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20620 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20621 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20622 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20623 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20624 VMULOSW): New enum "unspec" values.
20625 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20626 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20627 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20628 altivec_vmulosw): New patterns.
20629 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20630 VMULOSW): Add definitions.
20632 2017-06-20 Julia Koval <julia.koval@intel.com>
20634 * config/i386/i386.c: Fix rounding expand for new pattern.
20635 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20637 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20639 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20640 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20642 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20644 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20647 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20649 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20650 architecture, then by implementer ID.
20651 * config/aarch64/aarch64-tune.md: Regenerate.
20653 2017-06-20 Richard Biener <rguenther@suse.de>
20655 PR middle-end/81097
20656 * fold-const.c (split_tree): Fold to type before negating.
20658 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20660 * diagnostic-show-locus.c
20661 (selftest::test_fixit_deletion_affecting_newline): New function.
20662 (selftest::diagnostic_show_locus_c_tests): Call it.
20664 2017-06-20 Andreas Schwab <schwab@suse.de>
20667 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20670 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20672 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20673 __ARM_FEATURE_COPROC according to support.
20675 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20677 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20678 Rewritten to avoid overflow for > 32-bit pointers.
20681 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20682 by removing enum keyword.
20683 (ubsan_type_descriptor): Likewise. Formatting fix.
20686 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20687 splitter): Require TARGET_SSE2 in the condition.
20689 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20692 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20693 for doing vector set of SFmode on ISA 3.0.
20694 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20695 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20697 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20698 SFmode value into a V4SF variable that was extracted from another
20699 V4SF variable without converting the element to double precision
20700 and back to single precision vector format.
20701 (vsx_insert_extract_v4sf_p9_2): Likewise.
20703 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20705 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20706 in UWHI to avoid undefined overflow.
20709 * ubsan.h (enum ubsan_encode_value_phase): New.
20710 (ubsan_encode_value): Change second argument to
20711 enum ubsan_encode_value_phase with default value of
20712 UBSAN_ENCODE_VALUE_GENERIC.
20713 * ubsan.c (ubsan_encode_value): Change second argument to
20714 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20715 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20716 create_tmp_var_raw instead of create_tmp_var and use a
20718 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20719 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20720 ubsan_encode_value callers.
20723 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20724 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20725 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20727 2017-06-19 Richard Biener <rguenther@suse.de>
20729 PR middle-end/81118
20730 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20731 estimates if we changed anything.
20733 2017-06-19 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/80887
20736 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20737 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20738 simplified lookups, then reset mprts_hook.
20739 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20741 (try_to_simplify): Likewise.
20743 2017-06-19 Martin Liska <mliska@suse.cz>
20746 * gimplify.c (gimplify_switch_expr):
20747 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20749 2017-06-19 Martin Liska <mliska@suse.cz>
20751 * doc/install.texi: Document that PGO runs in 4 stages.
20753 2017-06-19 Martin Liska <mliska@suse.cz>
20756 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20757 to dispatcher function name.
20758 * multiple_target.c (replace_function_decl): New function.
20759 (create_dispatcher_calls): Redirect both edges and references.
20761 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20763 * profile-count.c (profile_count::dump): Dump quality.
20764 (profile_count::differs_from_p): Update for unsigned val.
20765 * profile-count.h (profile_count_quality): New enum.
20766 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20768 2017-06-19 Richard Biener <rguenther@suse.de>
20770 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20771 struct function as arg.
20772 (estimate_numbers_of_iterations): Export overload with loop arg.
20773 (free_numbers_of_iterations_estimates_loop): Use an overload of
20774 free_numbers_of_iterations_estimates instead.
20775 * tree-cfg.c (remove_bb): Adjust.
20776 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20777 * tree-parloops.c (gen_parallel_loop): Likewise.
20778 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20780 (tree_unroll_loops_completely): Likewise.
20781 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20782 Use an overload instead and export.
20783 (estimated_loop_iterations): Adjust.
20784 (max_loop_iterations): Likewise.
20785 (likely_max_loop_iterations): Likewise.
20786 (estimate_numbers_of_iterations): Take struct function as arg
20788 (loop_exits_before_overflow): Adjust.
20789 (free_numbers_of_iterations_estimates_loop): Use an overload.
20790 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20791 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20793 2017-06-19 Richard Biener <rguenther@suse.de>
20796 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20797 RANGE_EXPR conservatively.
20799 2017-06-16 Carl Love <cel@us.ibm.com>
20801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20802 definitions for vec_float, vec_float2, vec_floato,
20803 vec_floate built-ins.
20804 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20805 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20807 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20808 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20809 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20810 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20811 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20812 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20813 vec_floato): Add builtin defines.
20814 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20815 Update the built-in documentation file for the new built-in
20818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20820 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20821 (mthumb): Mark as the negative of -marm.
20823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20825 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20827 (ARM Options, -mtune): Document that this accepts the same
20828 extension options as -mcpu.
20829 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20833 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20834 permitted extensions.
20836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20838 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20839 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20840 (armv8-m.main): Add option +nodsp.
20841 * config/arm/arm-cpu-cdata.h: Regenerated.
20843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20845 * config/arm/t-fuchsia: New file.
20846 * config.gcc (arm*-*-fuchsia*): Use it.
20848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20850 * config/arm/t-symbian: Rewrite for new option infrastructure.
20852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20854 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20855 (MULTILIB_REQUIRED): Likewise.
20857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20859 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20860 (MULTILIB_RESUE): Likewise.
20861 (MULTILIB_MATCHES): Likewise.
20862 (MULTLIB_REQUIRED): Likewise.
20864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20866 * config/arm/t-rtems: Rewrite for new option framework.
20868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20870 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20871 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20872 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20873 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20874 * config/arm/t-multilib: ... here.
20875 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20876 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20877 armv7-a and armv8*-a when A-profile libraries have not been built.
20878 * config/arm/t-rmprofile: Rewrite.
20880 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20882 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20883 with a backslash. Remove the backslash after substituting unescaped
20885 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20889 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20890 the driver to pass through the default setting of -mfloat-abi.
20891 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20893 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20894 (all_feat_combs): New rule.
20895 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20897 * config/arm/t-aprofile: Rewrite.
20899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20901 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20902 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20903 fpu is not specified by the user/command-line.
20904 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20905 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20906 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20907 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20908 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20909 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20911 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20913 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20914 * config/arm/t-arm-elf: Rewritten.
20916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20918 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20919 have some floating-point instructions.
20920 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20921 (TARGET_MAYBE_HARD_FLOAT): New macro.
20922 * config/arm/arm-builtins.c (arm_init_builtins): Use
20923 TARGET_MAYBE_HARD_FLOAT.
20924 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20926 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20928 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20929 (configargs.h): Include it.
20930 (arm_print_hint_for_fpu_option): New function.
20931 (arm_parse_fpu_option): New function.
20932 (candidate_extension): New class.
20933 (arm_canon_for_multilib): New function.
20934 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20935 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20936 (ARCH_CANONICAL_SPECS): New macro.
20937 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20941 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20942 are set after handling multilib fragments. Set target_cpu_default2
20945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20947 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20949 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20950 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20952 (check_arch): Likewise.
20953 * config/arm/arm.c (arm_configure_build_target): Handle
20954 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20955 options in the default.
20957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20959 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20960 when an option is an alias of another.
20961 * config/arm/parsecpu.awk (optalias): New parser token.
20962 (gen_comm_data): Mark non-alias options as such. Emit entries
20963 for extension aliases.
20964 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20965 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20966 (armv6kz, armv6zk, armv6t2): Likewise.
20967 (armv7): Make vfpv3-d16 an alias.
20968 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20970 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20971 Sort in canonical order.
20972 (armv8-a): Sort in canonical order.
20973 (armv8.1-a, armv8.2-a): Likewise.
20974 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20976 (cortex-a9): Sort in canonical order.
20977 * config/arm/arm.c (selftests.h): Include it.
20978 (arm_test_cpu_arch_data): New function.
20979 (arm_run_self_tests): New function.
20980 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20981 (targetm): Move declaration to the end of the file.
20982 * arm-cpu-cdata.h: Regenerated.
20984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20986 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20987 call to target_mode_check describing the type of option passed.
20988 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20989 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20990 arm_parse_cpu_option_name to match parameters against list of
20992 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20993 arm_arch_core_flags data structure.
20994 * config/arm/arm-cpu_cdata.h: Regenerated.
20996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20998 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21000 (arm_print_hint_for_cpu_option): Likewise.
21001 (arm_print_hint_for_arch_option): Likewise.
21002 (arm_parse_cpu_option_name): Likewise.
21003 (arm_parse_arch_option_name): Likewise.
21004 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21005 of entries in the all_fpus list.
21006 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21007 (arm_parse_cpu_option_name): Declare.
21008 (arm_parse_arch_option_name): Declare.
21009 (arm_parse_option_features): Declare.
21010 (arm_intialize_isa): Declare.
21011 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21013 (gen_comm_data): ... here. Make definitions non-static.
21014 * config/arm/arm-cpu-data.h: Regenerated.
21015 * config/arm/arm-cpu-cdata.h: Regenerated.
21017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21019 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21020 (cpu_arch_extension): New structure.
21021 (cpu_arch_option, arch_option, cpu_option): New structures.
21022 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21023 architecture types.
21024 (gen_data): Generate new format data tables.
21025 * config/arm/arm.c (cpu_tune): New structure.
21026 (cpu_option, processors): Delete.
21027 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21028 (arm_print_hint_for_cpu_option): ... this and ...
21029 (arm_print_hint_for_arch_option): ... this.
21030 (arm_parse_arch_cpu_name): Delete. Replace with ...
21031 (arm_parse_cpu_option_name): ... this and ...
21032 (arm_parse_arch_option_name): ... this.
21033 (arm_unrecognized_feature): Change type of target parameter to
21035 (arm_parse_arch_cpu_features): Delete. Replace with ...
21036 (arm_parse_option_features): ... this.
21037 (arm_configure_build_target): Rework to use new configuration data
21039 (arm_print_tune_info): Rework for new configuration data tables.
21040 * config/arm/arm-cpu-data.h: Regenerated.
21041 * config/arm/arm-cpu.h: Regenerated.
21043 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21045 * Makefile.in (OBJS): Move sbitmap.o from here ...
21046 (OBJS-libcommon): ... to here.
21048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21050 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21051 (ISA_ALL_CRYPTO): New macro.
21052 (ISA_ALL_SIMD): New macro
21053 (ISA_ALL_FP): New macro.
21054 * config/arm/arm.c (fpu_bitlist): Update initializer.
21055 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21057 (arm9e): Add fpu. Add option for nofp
21058 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21059 (arm926ej-s, arm1026ej-s): Likewise.
21060 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21061 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21062 neon-fp16, neon-vfpv4, nofp and nosimd.
21063 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21064 (cortex-a8): Add fpu. Add option for nofp.
21065 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21066 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21067 (cortex-r4f): Add fpu.
21068 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21069 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21071 (cortex-r8): Likewise.
21072 (cortex-m4): Add fpu. Add option for nofp.
21073 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21074 (cortex-a17.cortex-a7): Likewise.
21075 (cortex-a32): Add fpu. Add options for crypto and nofp.
21076 (cortex-a35, cortex-a53): Likewise.
21077 (cortex-a57): Add fpu. Add option for crypto.
21078 (cortex-a72, cortex-a73): Likewise.
21079 (exynos-m1): Likewise.
21080 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21081 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21082 (cortex-m33): Add fpu. Add option for nofp.
21083 * config/arm/arm-cpu-cdata.h: Regenerated
21084 * config/arm/arm-cpu-data.h: Regenerated.
21086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21088 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21089 (armv5te, armv5tej): Likewise.
21090 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21091 (armv7): Add options fp and vfpv3-d16.
21092 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21093 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21095 (armv7ve): Likewise.
21096 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21097 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21098 (armv8-a): Add nocrypto option.
21099 (armv8.1-a, armv8.2-a): Likewise.
21100 (armv8-m.main): add options fp, fp.dp and nofp.
21102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21104 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21106 (armv8-a+crc): Delete.
21107 (armv8.1-a): Add options simd, crypto and nofp.
21108 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21109 (armv8.2-a+fp16): Delete.
21110 (armv8-m.main): Add option dsp.
21111 (armv8-m.main+dsp): Delete.
21112 (cortex-a8): Add fpu. Add nofp option.
21113 (cortex-a9): Add fpu. Add nofp and nosimd options.
21114 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21115 link to main cpu and architecture data structures.
21116 (gen_comm_data): Only put isa attributes from the main architecture
21118 (option): New statement for architecture and CPU entries.
21119 * arm.c (struct cpu_option): New structure.
21120 (struct processors): Add entry for options.
21121 (arm_unrecognized_feature): New function.
21122 (arm_parse_arch_cpu_name): Ignore any characters after the first
21124 (arm_parse_arch_cpu_feature): New function.
21125 (arm_configure_build_target): Separate out any CPU and architecture
21126 features and parse separately. Don't error out if -mfpu=auto is
21127 used with only an architecture string.
21128 (arm_print_asm_arch_directives): New function.
21129 (arm_file_start): Call it.
21130 * config/arm/arm-cpu-cdata.h: Regenerated.
21131 * config/arm/arm-cpu-data.h: Likewise.
21132 * config/arm/arm-tables.opt: Likewise.
21134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21136 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21137 assembler when it is not -mfpu=auto.
21139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21141 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21142 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21143 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21144 (ASM_CPU_SPEC): Rewrite.
21145 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21146 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21147 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21148 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21149 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21150 copied string is NUL-terminated. Also strip any characters prefixed
21152 (arm_rewrite_selected_arch): New function.
21153 (arm_rewrite_march): New function.
21155 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21157 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21158 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21159 (march, mcpu, mtune): Convert to string-based options.
21160 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21161 (arm_parse_arch_cpu_name): New function.
21162 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21163 identify selected architecture or CPU.
21164 (arm_option_save): New function.
21165 (TARGET_OPTION_SAVE): Redefine.
21166 (arm_option_restore): Restore string options.
21167 (arm_option_print): Print string options.
21169 2017-06-16 Martin Sebor <msebor@redhat.com>
21171 PR tree-optimization/80933
21172 PR tree-optimization/80934
21173 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21174 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21175 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21176 (gimple_fold_builtin): Call them.
21178 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21180 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21181 as unlikely; update profile.
21183 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21185 * predict.c (force_edge_cold): Handle declaring edges impossible
21188 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21190 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21192 (try_unroll_loop_completely): Fix reporting.
21194 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21196 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21198 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21201 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21202 if given a non-constant argument for an intrinsic which requires a
21205 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21207 * profile.c (compare_freqs): New function.
21208 (branch_prob): Sort edge list.
21209 (find_spanning_tree): Assume that the list is priority sorted.
21211 2017-06-16 Richard Biener <rguenther@suse.de>
21213 PR tree-optimization/81090
21214 * passes.def (pass_record_bounds): Remove.
21215 * tree-pass.h (make_pass_record_bounds): Likewise.
21216 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21217 make_pass_record_bounds): Likewise.
21218 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21219 not free niter estimates at the beginning but at the end.
21220 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21222 2017-06-16 Richard Biener <rguenther@suse.de>
21224 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21225 initializer to workaround ICE in host GCC 4.8.
21227 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21229 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21231 (clone_inlined_nodes): Update.
21233 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21235 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21236 prefetch settings, and enable prefetching by default at -O3.
21238 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21240 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21241 Set flag_prefetch_loop_arrays according to tuning data.
21243 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21245 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21246 New tune structure.
21247 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21248 [Unrelated to main purpose of the patch] Place the pointer field last
21249 to enable type checking errors when tune structure are wrongly merged.
21250 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21251 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21252 (thunderx2t99_prefetch_tune): New tune constants.
21253 (tune_params *_tunings): Update all tunings (no functional change).
21254 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21255 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21256 from tunings structures.
21258 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21261 * ubsan.c (instrument_null): Add T argument, use it instead
21262 of computing it based on IS_LHS.
21263 (instrument_object_size): Likewise.
21264 (pass_ubsan::execute): Adjust instrument_null and
21265 instrument_object_size callers to pass gimple_get_lhs or
21266 gimple_assign_rhs1 result to it. Use instrument_null instead of
21267 calling get_base_address and instrument_mem_ref. Handle
21268 aggregate call arguments for object-size sanitization.
21270 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21272 PR tree-optimization/81089
21273 * tree-vrp.c (is_masked_range_test): Validate operands of
21276 2017-06-15 Martin Sebor <msebor@redhat.com>
21279 * dumpfile.c (dump_register): Avoid calling memset to initialize
21280 a class with a default ctor.
21281 * gcc.c (struct compiler): Remove const qualification.
21282 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21283 * hash-table.h: Ditto.
21284 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21286 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21287 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21289 * params.h (struct param_info): Make struct members non-const.
21290 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21291 with default initialization.
21292 * vec.h (vec_copy_construct, vec_default_construct): New helper
21294 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21295 with vec_copy_construct.
21296 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21297 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21298 * doc/invoke.texi (-Wclass-memaccess): Document.
21300 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21302 * emit-rtl.h (is_leaf): Update comment about local
21303 register allocator.
21305 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21308 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21309 for a variable to have a section before checking if the section has a
21311 Set section to.persistent if persistent attribute is set.
21312 Warn if .persistent attribute is used on an automatic variable.
21314 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21316 PR rtl-optimization/80474
21317 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21319 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21321 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21324 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21326 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21327 (casesi): Emit bounds checking as RTL.
21328 (casesi_internal_mips16_<mode>): Remove bounds checking.
21330 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21332 * config/xtensa/xtensa.c (xtensa_option_override): Append
21333 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21334 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21335 xtensa_doloop_hooks): Define unconditionally.
21336 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21338 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21339 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21340 for it in xtensa_option_override.
21341 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21342 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21344 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21346 * doc/cppopts.texi: Document '-' special value to -MF.
21348 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21350 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21351 (cortex_a53_fconst): Likewise.
21352 (cortex_a53_fpmul): Likewise.
21353 (cortex_a53_f_load_64): Likewise.
21354 (cortex_a53_f_load_many): Likewise.
21355 (cortex_a53_advsimd_alu): Likewise.
21356 (cortex_a53_advsimd_alu_q): Likewise.
21357 (cortex_a53_advsimd_mul): Likewise.
21358 (cortex_a53_advsimd_mul_q): Likewise.
21359 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21360 Add missing fmul, r2f_cvt and fconst cases.
21362 2017-06-14 Richard Biener <rguenther@suse.de>
21364 PR middle-end/81088
21365 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21367 (fold_binary_loc): When associating do not treat pre-existing
21368 TREE_OVERFLOW on literal constants as a reason to allow
21369 TREE_OVERFLOW on associated literal constants.
21371 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21373 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21374 (MASK_FEATURES): New macro.
21375 * config/sparc/sparc.c (sparc_option_override): Remove the special
21376 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21378 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21380 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21381 a division of 0 if non-call exceptions are enabled.
21383 2017-06-14 Andrew Pinski <apinski@cavium.com>
21384 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21387 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21388 Improve vector initialization code gen for only variable case.
21390 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21392 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21394 2017-06-14 Richard Biener <rguenther@suse.de>
21396 PR tree-optimization/81083
21397 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21400 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21402 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21403 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21404 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21405 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21406 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21407 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21409 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21411 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21412 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21414 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21416 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21418 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21420 * config/rs6000/t-rtems: Don't handle SPE.
21422 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21424 * config/rs6000/t-linux: Don't handle SPE.
21426 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21428 * config/rs6000/eabispe.h: Delete file.
21430 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21432 * config/rs6000/t-spe: Delete file.
21434 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21436 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21437 (rs6000_legitimate_offset_address_p): Return false for anything in
21438 V2SImode or V2SFmode.
21440 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21442 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21443 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21444 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21446 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21447 (rs6000_legitimate_offset_address_p): Ditto.
21448 (rs6000_emit_move): Ditto.
21449 (rs6000_init_builtins): Remove V4HI_type_node.
21451 2017-06-13 Martin Liska <mliska@suse.cz>
21454 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21455 (gate_asan): Likewise.
21456 * asan.h (asan_no_sanitize_address_p): Remove the function.
21457 (sanitize_flags_p): New function.
21458 * builtins.def: Fix coding style.
21459 * common.opt: Use renamed enum value.
21460 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21461 * doc/extend.texi: Document no_sanitize attribute.
21462 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21463 to SANITIZE_UNDEFINED_NONDEFAULT.
21464 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21465 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21466 Use sanitize_flags_p.
21467 * gimplify.c (gimplify_function_tree): Likewise.
21468 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21469 * opts.c (parse_no_sanitize_attribute): New function.
21470 (common_handle_option): Use renamed enum value.
21471 * opts.h (parse_no_sanitize_attribute): Declare.
21472 * tree.c (sanitize_flags_p): New function.
21473 * tree.h: Declared here.
21474 * tsan.c: Use sanitize_flags_p.
21475 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21476 (instrument_mem_ref): Likewise.
21477 (instrument_bool_enum_load): Likewise.
21478 (do_ubsan_in_current_function): Remove the function.
21479 (pass_ubsan::execute): Use sanitize_flags_p.
21480 * ubsan.h: Remove do_ubsan_in_current_function
21481 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21482 (dump_function_to_file): Use it here.
21484 2017-06-13 Martin Jambor <mjambor@suse.cz>
21486 PR tree-optimization/80803
21487 PR tree-optimization/81063
21488 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21489 (propagate_subaccesses_across_link): Enqueue subtree whenever
21490 necessary instead of relying on the caller.
21492 2017-06-13 Martin Jambor <mjambor@suse.cz>
21494 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21495 that have a first_link.
21496 (sort_and_splice_var_accesses): Do not check first_link before
21498 (subtree_mark_written_and_enqueue): Likewise.
21499 (propagate_all_subaccesses): Likewise and do not stop at first
21500 parent with a first_link.
21502 2017-06-13 Martin Jambor <mjambor@suse.cz>
21504 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21507 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21509 * match.pd: New pattern.
21511 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21513 * tree-vrp.c (is_masked_range_test): New function.
21514 (register_edge_assert_for): Determine ranges for
21517 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21519 PR tree-optimization/67328
21520 * fold-const.c (maskable_range_p): New function.
21521 (build_range_check): Generate bittests if possible.
21523 2017-06-13 Martin Liska <mliska@suse.cz>
21525 * gimple-pretty-print.c (dump_probability): Add new argument.
21526 (dump_edge_probability): Dump both probability and count.
21527 (dump_gimple_label): Likewise.
21528 (dump_gimple_bb_header): Likewise.
21530 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21533 * config/avr/avr-devices.c: Fix indentation.
21534 * config/avr/gen-avr-mmcu-specs.c: Dito.
21536 2017-06-13 Richard Biener <rguenther@suse.de>
21538 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21539 instead get vector type from stmt_info.
21540 (vectorizable_reduction): Adjust. Remove dead code.
21542 2017-06-13 Richard Biener <rguenther@suse.de>
21544 PR middle-end/81065
21545 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21546 case of C * (x * C2 + C3).
21547 (fold_addr_of_array_ref_difference): Properly fold index difference.
21549 2017-06-12 David S. Miller <davem@davemloft.net>
21552 * config/sparc/sparc.md (return expander): Emit frame blockage if
21553 function uses alloca.
21555 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21557 * combine.c (make_field_assignment): Check len rather than the mode
21558 precision when calling force_to_mode.
21560 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21562 Support multilibs and devices that see flash in RAM address range.
21565 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21566 (avr_mcu_t) <flash_pm_offset>: New field.
21567 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21568 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21569 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21570 (AVR_TINY_PM_OFFSET): Remove macro.
21571 * config/avr/avr.opt (-mshort-calls): New option.
21572 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21573 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21574 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21575 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21576 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21577 instead of avr_arch->have_jmp_call.
21578 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21579 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21580 avr_arch->flash_pm_offset to define.
21581 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21582 new field flash_pm_offset. Add entry for avrxmega3.
21583 (avr_texinfo): Add entry for avrxmega3.
21584 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21585 attiny212, attiny214,
21586 attiny412, attiny414, attiny416, attiny417,
21587 attiny814, attiny816, attiny817,
21588 attiny1614, attiny1616, attiny1617,
21589 attiny3214, attiny3216, attiny3217.
21590 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21591 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21592 (avr_print_operand_address) [AVR_TINY]: Same.
21593 (avr_asm_init_sections) <readonly_data_section>: Only patch
21594 callback if avr_arch->flash_pm_offset = 0.
21595 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21596 for rodata if avr_arch->flash_pm_offset != 0.
21597 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21598 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21599 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21600 (m_options): Append opt_rcall.
21601 (m_dirnames): Append dir_rcall.
21602 * config/avr/t-multilib: Regenerate.
21604 * configure.ac [target=avr]: Check whether avrxmega3 default
21605 linker description file works as needed.
21606 * configure: Regenerate.
21607 * doc/avr-mmcu.texi: Regenerate.
21608 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21609 <__AVR_ARCH__>: Document avrxmega3 and 103.
21610 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21611 <__AVR_SHORT_CALLS__>: Document it.
21612 <__AVR_PM_BASE_ADDRESS__>: Document it.
21613 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21614 (AVR Variable Attributes) <progmem>: Document this is
21615 not needed for avrxmega3.
21616 (AVR Named Address Spaces) <__flash>: Dito.
21618 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21620 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21622 2017-06-12 Doug Rupp <rupp@adacore.com>
21624 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21625 Append vxworks-stdint.h to the tm_file list.
21626 * config/vxworks-stdint.h: New file.
21628 2017-06-12 Martin Liska <mliska@suse.cz>
21630 PR tree-optimization/81041
21631 * tree-profile.c (gimple_gen_ic_func_profiler):
21632 Create an extra BB in profile-generate
21633 (gimple_gen_time_profiler): Likewise.
21635 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21637 PR tree-optimization/81003
21638 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21639 (update_range_test): Use it instead of force_gimple_operand_gsi.
21641 2017-06-12 Richard Biener <rguenther@suse.de>
21643 PR tree-optimization/81053
21644 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21645 with backedge value not defined in loop. Simplify def stmt
21648 2017-06-11 Tom de Vries <tom@codesourcery.com>
21651 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21653 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21654 nvptx_cannot_force_const_mem.
21656 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21658 * opts.c (finish_options): Move test for flag_split_stack after
21659 it has been initialized.
21661 2017-06-11 Jason Merrill <jason@redhat.com>
21663 * tree.h (id_equal): New.
21664 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21665 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21666 instead of strcmp of IDENTIFIER_POINTER.
21668 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21670 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21671 (mark_all_inlined_calls_cdtor): Fix formating.
21672 (inline_transform): Rescale profile before inlining.
21674 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21676 * cgraph.h (cgraph_edge::clone): Update prototype.
21677 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21678 (cgraph_node::create_clone): Update.
21679 (cgraph_node::create_version_clone): Update.
21680 * tree-inline.c (copy_bb): Update.
21681 (expand_call_inline): Update.
21683 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21685 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21686 factored out from ...
21687 (rs6000_emit_prologue): ... here.
21689 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21691 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21692 factored out from ...
21693 (rs6000_emit_prologue): ... here.
21695 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21697 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21699 (handle_missing_profiles): Fix computation of tp_first_run.
21700 (counts_to_freqs): Do not touch freqs when count is 0.
21702 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21704 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21707 2017-06-10 Tom de Vries <tom@codesourcery.com>
21709 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21710 attributes): Document signal effective target.
21712 2017-06-10 Tom de Vries <tom@codesourcery.com>
21714 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21715 Document effective target stack_size.
21717 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21719 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21720 to the edit_context if they can be auto-applied.
21722 2017-06-9 Ian Lance Taylor <iant@golang.org>
21724 * opts.c (finish_options): If -fsplit-stack, disable implicit
21725 -forder-blocks-and-partition.
21726 * doc/invoke.texi (Optimize Options): Document that when using
21727 -fsplit-stack -forder-blocks-and-partition is not implicitly
21730 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21732 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21733 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21734 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21735 * builtins.def (abort, trap, unreachable): Declare cold.
21736 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21737 * tree-core.h (ECF_COLD): New.
21738 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21739 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21741 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21743 * predict.c (unlikely_executed_stmt_p): Cleanup.
21745 2017-06-09 Richard Biener <rguenther@suse.de>
21747 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21748 model if the ref is always written to.
21750 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21752 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21754 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21756 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21759 2017-06-09 Tom de Vries <tom@codesourcery.com>
21762 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21763 "target cannot support label values" when encountering LABEL_REF.
21765 2017-06-09 Martin Liska <mliska@suse.cz>
21767 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21768 (gimple_gen_ic_func_profiler): Emit direct comparison
21769 of __gcov_indirect_call_callee with NULL.
21770 (gimple_gen_time_profiler): Change probability from
21771 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21773 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21775 * profile.c (edge_gcov_counts): Turn to pointer.
21776 (compute_branch_probabilities, compute_branch_probabilities): Update.
21777 (branch_prob): Do not clear edge_gcov_count.
21778 * profile.h (edge_gcov_counts): Turn to pointer.
21779 (edge_gcov_count): Update.
21781 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21783 * gimple.h (gimple_check_failed): Mark cold.
21785 2017-06-09 Richard Biener <rguenther@suse.de>
21787 PR tree-optimization/66623
21788 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21789 refactor check_reduction into two parts, properly computing
21790 whether we have to check reduction validity for outer loop
21793 2017-06-09 Richard Biener <rguenther@suse.de>
21795 PR tree-optimization/79483
21796 * graphite-scop-detection.c (order): New global.
21797 (get_order): Compute bb to order mapping that satisfies code
21798 generation constraints.
21799 (cmp_pbbs): New helper.
21800 (build_scops): Start domwalk at entry block, sort generated
21803 2017-06-09 Richard Biener <rguenther@suse.de>
21805 PR middle-end/81007
21806 * ipa-polymorphic-call.c
21807 (ipa_polymorphic_call_context::restrict_to_inner_class):
21808 Skip FIELD_DECLs with error_mark_node type.
21809 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21812 2017-06-09 Martin Liska <mliska@suse.cz>
21814 * predict.c (struct branch_predictor): New struct.
21815 (test_prediction_value_range): New test.
21816 (predict_c_tests): New function.
21817 * selftest-run-tests.c (selftest::run_tests): Run the function.
21818 * selftest.h: Declare new tests.
21820 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21823 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21824 gen_add3_insn did not fail.
21825 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21826 r0, construct that number in a temporary reg and add that reg to r0.
21827 If asked to put the result in r0 as well, fail.
21829 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21831 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21832 for early expansion of vec_eqv.
21834 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21836 PR middle-end/81005
21837 * ubsan.c (instrument_null): Avoid pointless code temporary.
21838 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21840 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21844 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21847 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21850 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21852 * predict.c (unlikely_executed_edge_p): Move ahead.
21853 (probably_never_executed_edge_p): Use it.
21855 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21857 PR middle-end/79988
21858 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21859 gimple_call_builtin_p call.
21861 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21863 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21864 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21865 rtl_check_failed_type2, rtl_check_failed_code1,
21866 rtl_check_failed_code2, rtl_check_failed_code_mode,
21867 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21868 rtvec_check_failed_bounds, rtl_check_failed_flag,
21869 _fatal_insn_not_found, _fatal_insn): Likewise.
21870 * tree.h (tree_contains_struct_check_failed,
21871 tree_check_failed, tree_not_check_failed,
21872 tree_class_check_failed, tree_range_check_failed,
21873 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21874 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21875 tree_operand_check_failed, omp_clause_check_failed,
21876 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21879 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21881 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21882 flag_branch_probabilities.
21883 * ipa-inline.c (edge_badness): Likewise.
21884 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21885 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21886 * predict.c (maybe_hot_frequency_p): Likewise.
21887 (probably_never_executed): Likewise.
21888 * sched-ebb.c (schedule_ebbs): Likewise.
21889 * sched-rgn.c (find_single_block_region): Likewise.
21890 * tracer.c (tail_duplicate): Likewise.
21892 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21894 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21895 longer requires x_flag_profile_use.
21897 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21899 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21900 instead of flag_reorder_blocks_and_partition.
21901 * dbxout.c (dbxout_function_end): Likewise.
21902 * dwarf2out.c (gen_subprogram_die): Likewise.
21903 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21904 * hw-doloop.c (reorg_loops): Likewise.
21905 * varasm.c (assemble_start_function,
21906 assemble_end_function): Likewise.
21907 (decide_function_section): Do not check for
21908 flag_reorder_blocks_and_partition.
21910 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21912 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21914 (chkp_get_hard_register_fake_addr_expr): Ditto.
21915 (chkp_build_addr_expr): Add check for hard reg case.
21916 (chkp_parse_array_and_component_ref): Ditto.
21917 (chkp_find_bounds_1): Ditto.
21918 (chkp_process_stmt): Don't generate bounds store for
21921 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21923 * predict.c (maybe_hot_bb_p): Do not check profile status.
21924 (maybe_hot_edge_p): Likewise.
21925 (probably_never_executed): Check for zero counts even if profile
21927 (unlikely_executed_edge_p): New function.
21928 (unlikely_executed_stmt_p): New function.
21929 (unlikely_executed_bb_p): New function.
21930 (set_even_probabilities): Use unlikely predicates.
21931 (combine_predictions_for_bb): Likewise.
21932 (predict_paths_for_bb): Likewise.
21933 (predict_paths_leading_to_edge): Likewise.
21934 (determine_unlikely_bbs): New function.
21935 (estimate_bb_frequencies): Use it.
21936 (compute_function_frequency): Use zero counts even if profile is
21938 * profile-count.h: Fix typo.
21940 2017-08-08 Julia Koval <julia.koval@intel.com>
21942 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21943 _mm512_mask_cvtsepi16_storeu_epi8,
21944 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21945 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21946 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21947 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21948 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21949 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21950 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21951 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21952 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21953 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21954 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21955 __builtin_ia32_pmovuswb256mem_mask,
21956 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21957 __builtin_ia32_pmovwb512mem_mask): New builtins.
21959 2017-08-08 Julia Koval <julia.koval@intel.com>
21961 PR target/73350,80862
21962 * config/i386/subst.md (round): Fix round pattern.
21963 * config/i386/i386.c (ix86_erase_embedded_rounding):
21964 Fix erasing rounding for the fixed pattern.
21966 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21968 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21970 2017-06-08 Martin Liska <mliska@suse.cz>
21972 PR gcov-profile/80911
21973 * gcov.c (block_info::block_info): New constructor.
21975 2017-06-07 Carl Love <cel@us.ibm.com>
21977 * config/rs6000/rs6000-c: The return type of the following
21978 built-in functions was implemented as int not long long. Fix sign
21979 of return value for the unsigned version of vec_mulo and vec_mule.
21980 vector unsigned long long vec_bperm (vector unsigned long long,
21981 vector unsigned char)
21982 vector signed long long vec_mule (vector signed int,
21984 vector unsigned long long vec_mule (vector unsigned int,
21985 vector unsigned int)
21986 vector signed long long vec_mulo (vector signed int,
21988 vector unsigned long long vec_mulo (vector unsigned int,
21989 vector unsigned int)
21990 * doc/extend.texi: Fix the documentation for the built-in
21993 2017-06-07 Carl Love <cel@us.ibm.com>
21996 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21999 2017-06-07 Carl Love <cel@us.ibm.com>
22001 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22002 support, Generate doublehv for signed int/float for BE case only.
22004 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22006 * doc/invoke.texi (mcx16): Rewrite.
22008 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22010 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22011 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22012 *mov<mode>_softfloat, and an anonymous splitter): Use
22013 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22015 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22017 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22019 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22020 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22021 (rs6000_debug_reg_global): Adjust.
22022 (rs6000_init_hard_regno_mode_ok): Adjust.
22023 (rs6000_dbx_register_number): Adjust.
22024 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22025 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22026 Remove SPE_ACC and SPEFSCR.
22027 (REG_ALLOC_ORDER): Ditto.
22028 (FRAME_POINTER_REGNUM): Change to 111.
22029 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22030 (REG_CLASS_NAMES): Ditto.
22031 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22032 (REGISTER_NAMES): Ditto.
22033 (ADDITIONAL_REG_NAMES): Ditto.
22034 (rs6000_reg_names): Ditto.
22035 * config/rs6000/rs6000.md: Renumber some register number
22038 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22040 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22042 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22043 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22045 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22046 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22047 Delete the SPE high registers.
22048 (REG_ALLOC_ORDER): Ditto.
22049 (enum reg_class): Remove SPE_HIGH_REGS.
22050 (REG_CLASS_NAMES): Ditto.
22051 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22052 (REGISTER_NAMES): Ditto.
22053 (rs6000_reg_names): Ditto.
22054 * doc/tm.texi.in: Remove SPE as example.
22055 * doc/tm.texi: Regenerate.
22057 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22059 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22060 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22061 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22062 * config/rs6000/rs6000.md (type): Remove "brinc".
22064 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22066 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22067 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22068 * config/rs6000/linuxspe.h: Delete file.
22069 * config/rs6000/rs6000.md: Don't include spe.md.
22070 * config/rs6000/spe.h: Delete file.
22071 * config/rs6000/spe.md: Delete file.
22072 * config/rs6000/t-rs6000: Remove spe.md.
22074 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22076 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22077 (reg_or_none500mem_operand): Delete.
22078 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22079 instead of reg_or_none500mem_operand.
22081 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22083 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22084 handling of SPE flags.
22085 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22087 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22089 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22091 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22092 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22093 paired_divv2sf3): Similar.
22094 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22095 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22096 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22098 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22099 Rename the paired_* instruction patterns.
22100 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22102 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22103 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22104 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22105 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22106 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22107 PAIRED_VECTOR_MODE.
22108 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22109 (spe_func_has_64bit_regs_p): Delete.
22110 (spe_expand_predicate_builtin): Delete.
22111 (spe_expand_evsel_builtin): Delete.
22112 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22113 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22114 (invalid_e500_subreg): Delete.
22115 (rs6000_legitimize_address): Always force_reg op2 as well, for
22116 paired single memory accesses.
22117 (rs6000_member_type_forces_blk): Delete.
22118 (rs6000_spe_function_arg): Delete.
22119 (rs6000_expand_unop_builtin): Delete SPE handling.
22120 (rs6000_expand_binop_builtin): Ditto.
22121 (spe_expand_stv_builtin): Delete.
22122 (bdesc_2arg_spe): Delete.
22123 (spe_expand_builtin): Delete.
22124 (spe_expand_predicate_builtin): Delete.
22125 (spe_expand_evsel_builtin): Delete.
22126 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22127 (spe_init_builtins): Delete.
22128 (spe_func_has_64bit_regs_p): Delete.
22129 (savres_routine_name): Delete "info" parameter. Adjust callers.
22130 (rs6000_emit_stack_reset): Ditto.
22131 (rs6000_dwarf_register_span): Delete.
22132 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22133 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22134 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22136 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22138 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22139 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22140 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22141 mulv2sf3, divv2sf3): Delete expanders.
22143 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22145 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22149 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22150 * config/rs6000/rs6000.c: Ditto.
22152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22154 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22155 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22156 comparison_operator.
22158 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22160 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22161 * config/rs6000/rs6000.opt: Ditto.
22162 * config/rs6000/t-rtems: Ditto.
22164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22166 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22167 TARGET_E500_SINGLE by 0, simplify.
22168 * config/rs6000/rs6000.c: Ditto.
22169 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22170 (spe_build_register_parallel): Delete.
22171 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22172 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22173 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22174 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22175 (E500_CONVERT): Delete.
22176 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22178 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22180 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22181 * config/rs6000/dfp.md: Ditto.
22182 (negdd2, *negdd2_fpr): Merge.
22183 (absdd2, *absdd2_fpr): Merge.
22184 (negtd2, *negtd2_fpr): Merge.
22185 (abstd2, *abstd2_fpr): Merge.
22186 * config/rs6000/e500.h: Delete file.
22187 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22188 TARGET_FPRS by 1 and simplify.
22189 * config/rs6000/rs6000-c.c: Ditto.
22190 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22191 TARGET_DF_SPE by 0.
22192 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22194 * config/rs6000/rs6000.md: Ditto.
22195 (floatdidf2, *floatdidf2_fpr): Merge.
22196 (move_from_CR_gt_bit): Delete.
22197 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22198 (E500_CR_IOR_COMPARE): Delete.
22199 (All patterns that require !TARGET_FPRS): Delete.
22200 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22202 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22204 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22206 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22208 * graphds.c (add_edge): Intitialize edge's attached data.
22209 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22210 pointer parameter. Call pointed function on each edge during
22211 graph traversing. Skip traversing the edge when the function
22213 (graphds_dfs, graphds_scc): Ditto.
22214 (for_each_edge): New parameter. Pass the new parameter to callback
22216 * graphds.h (skip_edge_callback): New function pointer type.
22217 (graphds_dfs, graphds_scc): New function pointer parameter.
22218 (graphds_edge_callback, for_each_edge): New parameter.
22220 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22222 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22223 out code checking if runtime alias check is possible to below ...
22224 Call the new function.
22225 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22226 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22228 2017-06-07 Marek Polacek <polacek@redhat.com>
22231 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22232 TYPE_OVERFLOW_WRAPS check.
22234 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22236 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22237 if versioning is required.
22238 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22239 peeling with the check for versioning.
22241 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22243 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22244 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22245 Set true to new parameter if new ssa variable is defined.
22246 (vect_gen_vector_loop_niters): Refactor. Set range information
22247 for the new vector loop bound variable.
22248 (vect_do_peeling): Ditto.
22250 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22252 * tree-affine.c (ssa.h): Include header file.
22253 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22254 has wrapping overflow behavior.
22256 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22258 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22260 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22262 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22263 (tree_to_aff_combination): ... here.
22265 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22267 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22268 reg_pressure model function.
22269 (ivopts_global_cost_for_size): Delete.
22270 (determine_set_costs, iv_ca_recount_cost): Call new model function
22271 ivopts_estimate_reg_pressure.
22273 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22275 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22276 expensive than udiv. Remove floating point cases from mod.
22278 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22280 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22281 Increase idiv cost.
22283 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22285 * config/aarch64/aarch64.md
22286 (copysignsf3): Fix mask generation.
22288 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22290 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22292 (class dump_manager): Add register_dumps method.
22293 * dumpfile.c: Include langhooks.h.
22294 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22295 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22296 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22297 (dump_manager::dump_register): Start with 512 entries instead of 32.
22298 (dump_manager::register_dumps): New method.
22299 * toplev.c (general_init): Instead of invoking register_dumps
22300 langhook, invoke register_dumps method on the dump manager.
22301 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22304 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22306 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22307 Say that # requires an associated define_split to exist, and that
22308 the define_split must be suitable for use after register allocation.
22310 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22312 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22313 (compute_outgoing_frequencies): Also initialize zero counts.
22314 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22315 around loops; preserve more of profile when nothing changes.
22317 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22319 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22321 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22322 * config/arm/arm-cpu-cdata.h: Regenerate.
22323 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22324 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22325 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22326 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22327 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22329 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22331 * config/arm/t-rmprofile: Likewise.
22332 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22334 2017-06-06 David S. Miller <davem@davemloft.net>
22337 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22338 blockage if function uses alloca.
22340 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22342 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22343 New "uid" fields to hold pretty-print IDs of group and ref.
22344 Memory references are now identified as <group_id>:<ref_id>
22345 instead of using [random] addresses.
22346 (dump_mem_details): Simplify, no functional change.
22347 (dump_mem_ref): Simplify and make output more concise.
22348 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22349 (find_or_create_group): Initialize group uid.
22350 (record_ref): Initialize ref uid. Improve debug output.
22351 (prune_group_by_reuse, should_issue_prefetch_p,)
22352 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22353 (mark_nontemporal_store, determine_loop_nest_reuse):
22354 Improve debug output.
22356 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22358 * dbgcnt.def (prefetch): New debug counter.
22359 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22360 (schedule_prefetches): Stop issueing prefetches if debug counter
22363 2017-06-06 Tom de Vries <tom@codesourcery.com>
22365 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22366 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22368 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22370 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22371 Use aarch64_reg_or_zero predicate for operand 4.
22372 (aarch64_compare_and_swap<mode> define_insn_and_split):
22373 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22374 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22376 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22378 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22379 (arm_compute_save_core_reg_mask): This.
22380 (thumb1_compute_save_reg_mask): Rename into ...
22381 (thumb1_compute_save_core_reg_mask): This.
22382 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22383 (arm_compute_frame_layout): Likewise.
22385 2017-06-06 Richard Biener <rguenther@suse.de>
22387 PR tree-optimization/80974
22388 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22389 keep or clear leaders SSA info.
22391 2017-06-06 Tom de Vries <tom@codesourcery.com>
22393 * config/nvptx/nvptx.c (split_mode_p): New function.
22394 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22396 2017-06-06 Tom de Vries <tom@codesourcery.com>
22398 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22400 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22403 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22406 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22408 * shrink-wrap.c (handle_simple_exit): Update profile.
22409 (try_shrink_wrapping): Upate profile.
22411 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22413 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22414 (tree_guess_outgoing_edge_probabilities): New.
22415 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22416 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22418 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22420 * ipa-split.c (split_function): Initialize return bb profile.
22422 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22424 * profile.c (compute_branch_probabilities): Also initialize
22425 EXIT_BLOCK profile.
22427 2017-06-06 Richard Biener <rguenther@suse.de>
22429 PR tree-optimization/80928
22430 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22431 (vect_analyze_loop_operations): Properly guard analysis for
22433 (vect_transform_loop): Likewise.
22434 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22435 (vect_model_induction_cost): Do not cost for pure SLP.
22436 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22437 of induction in inner loop vectorization.
22438 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22439 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22440 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22442 (vect_analyze_slp_cost_1): Cost induction.
22443 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22444 (vect_get_slp_vect_defs): Likewise.
22445 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22446 (vect_transform_stmt): Handle SLP reductions.
22447 * tree-vectorizer.h (vectorizable_induction): Adjust.
22449 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22451 * config/rs6000/rs6000.c (make_resolver_func): Update
22452 init_lowered_empty_function call.
22454 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22456 * doc/invoke.texi: Document the -fprofile-abs-path option.
22457 * common.opt (fprofile-abs-path): New option.
22458 * gcov-io.h (gcov_write_filename): Declare.
22459 * gcov-io.c (gcov_write_filename): New function.
22460 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22461 * profile.c (output_location): Likewise.
22463 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22465 * shring-wrap.c: Revert accidental commit.
22467 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22469 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22471 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22473 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22475 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22476 profile in callgraph edge.
22477 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22478 (apply_scale): Likewise.
22479 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22480 Also scale profile when inlining function with zero profile.
22481 (initialize_cfun): Update exit block profile even when it is zero.
22482 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22483 when profile is read.
22485 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22487 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22488 (CLONE_*): New constants to define the processors we can generate
22489 code for with the target_clone attribute.
22490 (rs6000_clone_map): New array to identify which clone processors
22491 the current program is running on.
22492 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22493 target_clone attribute.
22494 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22495 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22496 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22497 (cpu_expand_builtin): Add support for target_clone attribute.
22498 (rs6000_valid_attribute_p): Allow "default" attribute.
22499 (get_decl_name): New debug function to simplify printing the
22500 current function name in debugging statements.
22501 (rs6000_clone_priority): New functions to support the target_clone
22502 attribute, and be able to generate code to switch between ISA 2.05
22503 through ISA 3.0 (power6 through power9).
22504 (rs6000_compare_version_priority): Likewise.
22505 (rs6000_get_function_versions_dispatcher): Likewise.
22506 (make_resolver_func): Likewise.
22507 (add_condition_to_bb): Likewise.
22508 (dispatch_function_versions): Likewise.
22509 (rs6000_generate_version_dispatcher_body): Likewise.
22510 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22511 (fusion_gpr_load_p): Fix a spacing issue.
22512 * doc/extend.texi (Common Function Attributes): Document that the
22513 PowerPC supports the target_clone attribute.
22515 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22517 * config/arm/arm.h: explain F symbol found in description of ARM
22518 register allocation in its legend.
22520 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22522 * config/mips/frame-header-opt.c: Include profile-count.h.
22523 * config/riscv/riscv.c: Include profile-count.h
22525 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22527 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22529 (sm_set_flag_if_changed): Add bbs field.
22530 (execute_sm_if_changed_flag_set): Pass BBS.
22531 (execute_sm): Update.
22533 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22535 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22538 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22540 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22541 (peephole2): New peephole2 to emit the above.
22542 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22544 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22546 * config/aarch64/aarch64.c (define_peephole2 above
22547 *sub_<shift>_<mode>): New peephole.
22549 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22551 * config/i386/i386.c (make_resolver_func): Update.
22552 * Makefile.in: Add profile-count.h and profile-count.o
22553 * auto-profile.c (afdo_indirect_call): Update to new API.
22554 (afdo_set_bb_count): Update.
22555 (afdo_propagate_edge): Update.
22556 (afdo_propagate_circuit): Update.
22557 (afdo_calculate_branch_prob): Update.
22558 (afdo_annotate_cfg): Update.
22559 * basic-block.h: Include profile-count.h
22560 (struct edge_def): Turn count to profile_count.
22561 (struct basic_block_def): Likewie.
22562 (REG_BR_PROB_BASE): Move to profile-count.h
22563 (RDIV): Move to profile-count.h
22564 * bb-reorder.c (max_entry_count): Turn to profile_count.
22565 (find_traces): Update.
22566 (rotate_loop):Update.
22567 (connect_traces):Update.
22568 (sanitize_hot_paths):Update.
22569 * bt-load.c (migrate_btr_defs): Update.
22570 * cfg.c (RDIV): Remove.
22571 (init_flow): Use alloc_block.
22572 (alloc_block): Uninitialize count.
22573 (unchecked_make_edge): Uninitialize count.
22574 (check_bb_profile): Update.
22575 (dump_edge_info): Update.
22576 (dump_bb_info): Update.
22577 (update_bb_profile_for_threading): Update.
22578 (scale_bbs_frequencies_int): Update.
22579 (scale_bbs_frequencies_gcov_type): Update.
22580 (scale_bbs_frequencies_profile_count): New.
22581 * cfg.h (update_bb_profile_for_threading): Update.
22582 (scale_bbs_frequencies_profile_count): Declare.
22583 * cfgbuild.c (compute_outgoing_frequencies): Update.
22584 (find_many_sub_basic_blocks): Update.
22585 * cfgcleanup.c (try_forward_edges): Update.
22586 (try_crossjump_to_edge): Update.
22587 * cfgexpand.c (expand_gimple_tailcall): Update.
22588 (construct_exit_block): Update.
22589 * cfghooks.c (verify_flow_info): Update.
22590 (dump_bb_for_graph): Update.
22591 (split_edge): Update.
22592 (make_forwarder_block): Update.
22593 (duplicate_block): Update.
22594 (account_profile_record): Update.
22595 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22596 (get_estimated_loop_iterations): Update.
22597 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22598 (single_likely_exit): Update.
22599 * cfgloopmanip.c (scale_loop_profile): Update.
22601 (set_zero_probability): Update.
22602 (lv_adjust_loop_entry_edge): Update.
22603 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22604 (purge_dead_edges): Update.
22605 (rtl_account_profile_record): Update.
22606 * cgraph.c (cgraph_node::create): Uninitialize count.
22607 (symbol_table::create_edge): Uninitialize count.
22608 (cgraph_update_edges_for_call_stmt_node): Update.
22609 (cgraph_edge::dump_edge_flags): Update.
22610 (cgraph_node::dump): Update.
22611 (cgraph_edge::maybe_hot_p): Update.
22612 * cgraph.h: Include profile-count.h
22613 (create_clone), create_edge, create_indirect_edge): Update.
22614 (cgraph_node): Turn count to profile_count.
22615 (cgraph_edge0: Likewise.
22616 (make_speculative, clone): Update.
22617 (create_edge): Update.
22618 (init_lowered_empty_function): Update.
22619 * cgraphclones.c (cgraph_edge::clone): Update.
22620 (duplicate_thunk_for_node): Update.
22621 (cgraph_node::create_clone): Update.
22622 * cgraphunit.c (cgraph_node::analyze): Update.
22623 (cgraph_node::expand_thunk): Update.
22624 * final.c (dump_basic_block_info): Update.
22625 * gimple-streamer-in.c (input_bb): Update.
22626 * gimple-streamer-out.c (output_bb): Update.
22627 * graphite.c (print_global_statistics): Update.
22628 (print_graphite_scop_statistics): Update.
22629 * hsa-brig.c: Include basic-block.h.
22630 * hsa-dump.c: Include basic-block.h.
22631 * hsa-gen.c (T sum_slice): Update.
22632 (convert_switch_statements):Update.
22633 * hsa-regalloc.c: Include basic-block.h.
22634 * ipa-chkp.c (chkp_produce_thunks): Update.
22635 * ipa-cp.c (struct caller_statistics): Update.
22636 (init_caller_stats): Update.
22637 (gather_caller_stats): Update.
22638 (ipcp_cloning_candidate_p): Update.
22639 (good_cloning_opportunity_p): Update.
22640 (get_info_about_necessary_edges): Update.
22641 (dump_profile_updates): Update.
22642 (update_profiling_info): Update.
22643 (update_specialized_profile): Update.
22644 (perhaps_add_new_callers): Update.
22645 (decide_about_value): Update.
22646 (ipa_cp_c_finalize): Update.
22647 * ipa-devirt.c (struct odr_type_warn_count): Update.
22648 (struct decl_warn_count): Update.
22649 (struct final_warning_record): Update.
22650 (possible_polymorphic_call_targets): Update.
22651 (ipa_devirt): Update.
22652 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22653 * ipa-icf.c (sem_function::merge): Update.
22654 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22655 * ipa-inline.c (compute_uninlined_call_time): Update.
22656 (compute_inlined_call_time): Update.
22657 (want_inline_small_function_p): Update.
22658 (want_inline_self_recursive_call_p): Update.
22659 (edge_badness): Update.
22660 (lookup_recursive_calls): Update.
22661 (recursive_inlining): Update.
22662 (inline_small_functions): Update.
22663 (dump_overall_stats): Update.
22664 (dump_inline_stats): Update.
22665 * ipa-profile.c (ipa_profile_generate_summary): Update.
22666 (ipa_propagate_frequency): Update.
22667 (ipa_profile): Update.
22668 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22669 * ipa-utils.c (ipa_merge_profiles): Update.
22670 * loop-doloop.c (doloop_modify): Update.
22671 * loop-unroll.c (report_unroll): Update.
22672 (unroll_loop_runtime_iterations): Update.
22673 * lto-cgraph.c (lto_output_edge): Update.
22674 (lto_output_node): Update.
22675 (input_node): Update.
22676 (input_edge): Update.
22677 (merge_profile_summaries): Update.
22678 * lto-streamer-in.c (input_cfg): Update.
22679 * lto-streamer-out.c (output_cfg): Update.
22680 * mcf.c (create_fixup_graph): Update.
22681 (adjust_cfg_counts): Update.
22682 (sum_edge_counts): Update.
22683 * modulo-sched.c (sms_schedule): Update.
22684 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22685 * predict.c (maybe_hot_count_p): Update.
22686 (probably_never_executed): Update.
22687 (dump_prediction): Update.
22688 (combine_predictions_for_bb): Update.
22689 (propagate_freq): Update.
22690 (handle_missing_profiles): Update.
22691 (counts_to_freqs): Update.
22692 (rebuild_frequencies): Update.
22693 (force_edge_cold): Update.
22694 * predict.h: Include profile-count.h
22695 (maybe_hot_count_p, counts_to_freqs): UPdate.
22696 * print-rtl-function.c: Do not include cfg.h
22697 * print-rtl.c: Include basic-block.h
22698 * profile-count.c: New file.
22699 * profile-count.h: New file.
22700 * profile.c (is_edge_inconsistent): Update.
22701 (correct_negative_edge_counts): Update.
22702 (is_inconsistent): Update.
22703 (set_bb_counts): Update.
22704 (read_profile_edge_counts): Update.
22705 (compute_frequency_overlap): Update.
22706 (compute_branch_probabilities): Update; Initialize and deinitialize
22708 (branch_prob): Update.
22709 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22710 (edge_gcov_count): New.
22711 (bb_gcov_count): New.
22712 * shrink-wrap.c (try_shrink_wrapping): Update.
22713 * tracer.c (better_p): Update.
22714 * trans-mem.c (expand_transaction): Update.
22715 (ipa_tm_insert_irr_call): Update.
22716 (ipa_tm_insert_gettmclone_call): Update.
22717 * tree-call-cdce.c: Update.
22718 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22719 (gimple_duplicate_sese_tail): Update.
22720 (gimple_account_profile_record): Update.
22721 (execute_fixup_cfg): Update.
22722 * tree-inline.c (copy_bb): Update.
22723 (copy_edges_for_bb): Update.
22724 (initialize_cfun): Update.
22725 (freqs_to_counts): Update.
22726 (copy_cfg_body): Update.
22727 (expand_call_inline): Update.
22728 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22729 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22730 (try_unroll_loop_completely): Update.
22731 (try_peel_loop): Update.
22732 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22733 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22734 * tree-ssa-loop-split.c (connect_loops): Update.
22735 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22736 * tree-ssa-reassoc.c (branch_fixup): Update.
22737 * tree-ssa-tail-merge.c (replace_block_by): Update.
22738 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22739 (compute_path_counts): Update.
22740 (update_profile): Update.
22741 (recompute_probabilities): Update.
22742 (update_joiner_offpath_counts): Update.
22743 (estimated_freqs_path): Update.
22744 (freqs_to_counts_path): Update.
22745 (clear_counts_path): Update.
22746 (ssa_fix_duplicate_block_edges): Update.
22747 (duplicate_thread_path): Update.
22748 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22749 (struct switch_conv_info): Update.
22750 * tree-tailcall.c (decrease_profile): Update.
22751 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22752 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22753 * value-prof.c (check_counter): Update.
22754 (gimple_divmod_fixed_value): Update.
22755 (gimple_mod_pow2): Update.
22756 (gimple_mod_subtract): Update.
22757 (gimple_ic_transform): Update.
22758 (gimple_stringop_fixed_value): Update.
22759 * value-prof.h (gimple_ic): Update.
22761 2017-06-02 Carl Love <cel@us.ibm.com>
22763 * config/rs6000/rs6000-c: Add support for built-in functions
22764 vector double vec_doublee (vector signed int);
22765 vector double vec_doublee (vector unsigned int);
22766 vector double vec_doublee (vector float);
22767 vector double vec_doubleh (vector signed int);
22768 vector double vec_doubleh (vector unsigned int);
22769 vector double vec_doubleh (vector float);
22770 vector double vec_doublel (vector signed int);
22771 vector double vec_doublel (vector unsigned int);
22772 vector double vec_doublel (vector float);
22773 vector double vec_doubleo (vector signed int);
22774 vector double vec_doubleo (vector unsigned int);
22775 vector double vec_doubleo (vector float);.
22776 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22777 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22779 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22780 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22781 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22783 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22784 vec_doublel, vec_doubleh.
22785 * doc/extend.texi: Update the built-in documentation file for the
22786 new built-in functions.
22788 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22791 * ipa-inline-analysis.c (free_growth_caches): Set
22792 edge_removal_hook_holder to NULL after removing it.
22794 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22796 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22797 comparision with zero.
22799 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22800 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22801 for early expansion of vec_min and vec_max builtins.
22802 (builtin_function_type): Add min/max unsigned variants to those
22803 identified as having unsigned arguments.
22805 2017-06-02 Olivier Hainque <hainque@adacore.com>
22807 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22809 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22811 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22812 Use VALL_F16 iterator rather than VALL.
22814 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22816 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22817 Emit CBNZ inside loop when doing a strong exchange and comparing
22818 against zero. Generate the CC flags after the loop.
22820 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22822 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22823 (dl_section_ref): New.
22824 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22825 On AIX, append an expression to subtract the size of the
22826 section length to dl_section_ref.
22828 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22831 for early expansion of vector absolute builtins.
22833 2017-06-02 Richard Biener <rguenther@suse.de>
22835 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22836 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22838 2017-06-02 Richard Biener <rguenther@suse.de>
22840 PR tree-optimization/80948
22841 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22842 stmt order as well.
22844 2017-06-02 Richard Biener <rguenther@suse.de>
22846 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22848 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22849 for inductions as relevant.
22851 2017-06-02 Richard Biener <rguenther@suse.de>
22853 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22854 (vectorizable_induction): ... this. Remove dead code.
22856 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22858 * builtins. (expand_builtin_alloca): Remove second parameter and
22859 infer its value from the first parameter instead.
22860 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22862 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22864 PR rtl-optimization/80903
22865 * loop-doloop.c (add_test): Unshare sequence.
22867 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22869 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22871 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22873 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22875 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22876 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22877 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22878 (xlogue_layout::get_stub_rtx): Make static.
22879 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22880 (xlogue_layout::compute_stub_managed_regs): Rename to...
22881 (xlogue_layout::count_stub_managed_regs): ...this.
22882 (xlogue_layout::is_stub_managed_reg): New function.
22883 (xlogue_layout::m_stub_names): Rename to...
22884 (xlogue_layout::s_stub_names): ...this, make static.
22885 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22886 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22887 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22888 xlogue_layout::s_stub_names): Instantiate statics.
22889 (stub_managed_regs): Remove.
22890 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22891 (disable_call_ms2sysv_xlogues): Rename to...
22892 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22893 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22895 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22896 change after reload_completed.
22897 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22899 (ix86_expand_prologue): Likewise.
22900 (ix86_expand_epilogue): Likewise.
22901 (ix86_expand_split_stack_prologue): Likewise.
22902 (ix86_compute_frame_layout): Remove frame parameter ...
22903 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22904 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22906 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22907 (ix86_frame): Move from here ...
22908 * config/i386/i386.h (ix86_frame): ... to here.
22909 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22910 complete ix86_frame data structure instead. Remove some_ld_name.
22912 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22914 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22915 symbols that hold a DECL_VALUE_EXPR.
22917 2017-06-01 Martin Jambor <mjambor@suse.cz>
22919 PR tree-optimization/80898
22920 * tree-sra.c (process_subtree_disqualification): Removed.
22921 (disqualify_candidate): Do not acll
22922 process_subtree_disqualification.
22923 (subtree_mark_written_and_enqueue): New function.
22924 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22925 RHS has been disqualified and re-queue LHS if necessary. Apart
22926 from that, ignore disqualified RHS.
22928 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22930 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22931 address fetch for z10 or later.
22933 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22935 * config/arc/arc.md (tst_movb): Add guard when splitting.
22937 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22939 * config/arc/arc.c (arc_can_eliminate): Test against
22940 arc_frame_pointer_needed.
22942 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22944 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22945 to prevent store reordering.
22946 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22947 (type): Add block type.
22948 (stack_tie): Define special instruction to be used in
22951 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22953 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22954 constraint. It is not valid for the pattern.
22955 (noncommutative_binary_comparison): Likewise.
22957 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22959 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22962 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22964 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22965 be used by the reg-alloc.
22967 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22969 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22970 reg-alloc when having mul64 or mul32x16 instructions.
22971 (mulsidi3): Likewise.
22972 (umulsidi3): Likewise.
22973 (mulsi32x16): New pattern.
22974 (mulsi64): Likewise.
22975 (mulsidi64): Likewise.
22976 (umulsidi64): Likewise.
22977 (MUL32x16_REG): Define.
22978 (mul64_600): Use MUL32x16_REG.
22979 (mac64_600): Likewise.
22980 (umul64_600): Likewise.
22981 (umac64_600): Likewise.
22983 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22985 * config/arc/arc.md (mulsi3_700): Make it commutative.
22987 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22989 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22991 (*sign_extendsidi2_insn): Likewise for movstosw.
22993 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22995 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22996 the type of the input discriminant value. Convert the
22997 discriminant value of signedness vary.
22999 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23001 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23002 Add to -Wall section.
23004 2017-06-01 Richard Biener <rguenther@suse.de>
23006 PR middle-end/66313
23007 * fold-const.c (fold_plusminus_mult_expr): If the factored
23008 factor may be zero use a wrapping type for the inner operation.
23009 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23010 and handle moved defs.
23011 (process_assignment): Properly guard the unary op case. Return a
23012 tri-state indicating that moving the stmt before the call may allow
23013 to continue. Pass through to_move.
23014 (find_tail_calls): Handle moving unrelated defs before
23017 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23020 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23021 splitter result in the canonical way.
23023 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23025 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23026 also for 32bit target. Update insn attributes.
23027 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23029 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23031 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23032 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23033 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23034 (_mm_maskz_min_ss): New intrinsics.
23036 2017-05-31 Martin Liska <mliska@suse.cz>
23038 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23039 Change comment style to one we normally use.
23040 (vectorizable_reduction): Likewise.
23041 (vectorizable_induction): Likewise.
23042 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23043 (vectorizable_call): Likewise.
23044 (vectorizable_simd_clone_call): Likewise.
23045 (vectorizable_conversion): Likewise.
23046 (vectorizable_assignment): Likewise.
23047 (vectorizable_shift): Likewise.
23048 (vectorizable_operation): Likewise.
23049 (vectorizable_store): Likewise.
23050 (vectorizable_load): Likewise.
23051 * tree-vectorizer.h: Likewise.
23053 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23055 * passes.c (emergency_dump_function): New.
23056 * tree-pass.h (emergency_dump_function): Declare.
23057 * plugin.c (plugins_internal_error_function): Remove.
23058 * plugin.h (plugins_internal_error_function): Remove declaration.
23059 * toplev.c (internal_error_function): New static function. Use it...
23060 (general_init): ...here.
23062 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23064 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23065 (arc_rtx_costs): Add costs for new patterns.
23066 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23067 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23069 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23071 * tree-ssa-strlen.c (get_next_strinfo): New function.
23072 (get_stridx_plus_constant): Use it.
23073 (zero_length_string): Likewise.
23074 (adjust_related_strinfos): Likewise.
23075 (adjust_last_stmt): Likewise.
23077 2017-05-31 Richard Biener <rguenther@suse.de>
23080 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23081 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23083 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23085 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23086 loop_vinfo argument and use of dependence distance vectors.
23087 Check instead whether the two references differ only in their
23088 initial value and assume that they have the same alignment if the
23089 difference is a multiple of the vector alignment.
23090 (vect_analyze_data_refs_alignment): Update call accordingly.
23092 2017-05-31 Martin Liska <mliska@suse.cz>
23095 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23097 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23099 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23100 (create_intersect_range_checks): Move from ...
23101 * tree-data-ref.c (create_intersect_range_checks_index)
23102 (create_intersect_range_checks): ... to here.
23103 (create_runtime_alias_checks): New function factored from ...
23104 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23105 here. Call above function.
23106 * tree-data-ref.h (create_runtime_alias_checks): New function.
23108 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23110 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23111 segment length for dr_b and compute it in wide_int.
23113 2017-05-31 Richard Biener <rguenther@suse.de>
23115 PR tree-optimization/80906
23116 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23117 and pass through iv_map.
23118 (copy_bb_and_scalar_dependences): Adjust.
23119 (translate_pending_phi_nodes): Likewise.
23120 (copy_loop_close_phi_args): Handle code-generating IVs instead
23123 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23125 * diagnostic-color.c (color_dict): Add "type-diff".
23126 (parse_gcc_colors): Update comment.
23127 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23128 -fdiagnostics-show-template-tree and -fno-elide-type.
23129 (GCC_COLORS): Add type-diff to example.
23131 (-fdiagnostics-show-template-tree): New.
23132 (-fno-elide-type): New.
23133 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23134 the pp_format_decoder callback. Call any m_format_postprocessor's
23136 (pretty_printer::pretty_printer): Initialize
23137 m_format_postprocessor.
23138 (pretty_printer::~pretty_printer): Delete any
23139 m_format_postprocessor.
23140 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23141 (class format_postprocessor): New class.
23142 (struct pretty_printer::format_decoder): Document the new parameters.
23143 (struct pretty_printer::m_format_postprocessor): New field.
23144 * tree-diagnostic.c (default_tree_printer): Update for new
23145 bool and const char ** params.
23146 * tree-diagnostic.h (default_tree_printer): Likewise.
23148 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23150 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23151 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23152 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23153 rs6000_gen_cell_microcode code.
23154 (rs6000_final_prescan_insn): Delete.
23155 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23156 "warn-cell-microcode" entries.
23157 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23158 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23159 throughout. Change cc_reg_not_micro_cr0_operand to
23160 cc_reg_not_cr0_operand throughout.
23161 (*extendhi<mode>2_noload): Delete.
23162 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23163 (mwarn-cell-microcode): Delete.
23164 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23165 -mgen-cell-microcode and -mwarn-cell-microcode.
23167 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23170 * config/i386/constraints.md (Yd): New constraint.
23172 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23173 and (?Yd, r) alternatives. Update insn attributes.
23174 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23175 and (?*Yd, r) alternatives. Update insn attributes.
23176 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23178 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23180 * gimplify.c (gimplify_modify_expr): Don't create a
23181 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23184 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23186 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23188 2017-05-30 Richard Biener <rguenther@suse.de>
23190 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23191 and reduc_def fields.
23192 (STMT_VINFO_REDUC_TYPE): New define.
23193 (STMT_VINFO_REDUC_DEF): Likewise.
23194 (vect_force_simple_reduction): Adjust prototype.
23195 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23196 (vect_is_simple_reduction): Remove check_reduction argument.
23197 (vect_force_simple_reduction): Adjust and set
23198 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23199 (vectorizable_reduction): Do not re-do reduction analysis
23200 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23201 * tree-parloops.c (gather_scalar_reductions): Adjust.
23203 2017-05-30 Richard Biener <rguenther@suse.de>
23205 PR middle-end/80901
23206 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23209 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23211 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23212 Introduce unknown_misalignment parameter and remove vf.
23213 (vect_peeling_hash_get_lowest_cost):
23214 Pass unknown_misalignment parameter.
23215 (vect_enhance_data_refs_alignment):
23216 Fix unsupportable data ref treatment.
23218 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23220 * tree-vect-data-refs.c (vect_get_data_access_cost):
23221 Workaround for SLP handling.
23222 (vect_enhance_data_refs_alignment):
23223 Compute costs for doing no peeling at all, compare to the best
23224 peeling costs so far and avoid peeling if cheaper.
23226 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23228 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23229 Return peeling info and set costs to zero for unlimited cost
23231 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23232 with unknown misalignment. Compute and costs for unknown
23233 misalignment, compare them to the costs for known misalignment
23234 and choose the cheapest for peeling.
23236 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23238 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23239 (vect_get_peeling_costs_all_drs): Create function.
23240 (vect_peeling_hash_get_lowest_cost):
23241 Use vect_get_peeling_costs_all_drs.
23242 (vect_peeling_supportable): Create function.
23243 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23245 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23247 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23248 DR_HAS_NEGATIVE_STEP.
23249 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23250 (vect_enhance_data_refs_alignment): Use.
23251 (vect_duplicate_ssa_name_ptr_info): Use.
23252 * tree-vectorizer.h (dr_misalignment): Use.
23253 (known_alignment_for_access_p): Use.
23255 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23258 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23260 (has_section_name): New function.
23262 2017-05-30 Martin Liska <mliska@suse.cz>
23265 * auto-profile.c (get_function_decl_from_block): Fix
23268 2017-05-30 Richard Biener <rguenther@suse.de>
23270 PR middle-end/80876
23271 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23273 2017-05-30 Martin Liska <mliska@suse.cz>
23275 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23276 * dumpfile.h (struct dump_file_info): Remove ctors.
23278 2017-05-30 Martin Liska <mliska@suse.cz>
23280 * predict.def: Fix GNU coding style.
23282 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23284 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23285 Mark 'to' argument with ATTRIBUTE_UNUSED.
23287 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23289 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23290 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23291 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23294 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23296 * doc/install.texi (Options specification): Restore entry of
23297 --enable-sjlj-exceptions.
23299 2017-05-27 Michael Eager <eager@eagercon.com>
23302 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23304 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23306 * config/microblaze/microblaze.h
23307 (FIXED_REGISTERS): Update in macro.
23308 (CALL_USED_REGISTERS): Update in macro.
23310 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23312 * doc/install.texi: Add links to macOS binary distributions.
23314 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23318 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23320 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23322 2017-05-26 Martin Liska <mliska@suse.cz>
23324 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23326 2017-05-26 Martin Liska <mliska@suse.cz>
23328 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23329 always leading ';; '.
23330 (dump_bb_info): Likewise.
23331 (brief_dump_cfg): Likewise.
23332 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23333 * dumpfile.c: Remove usage of TDF_VERBOSE.
23334 * dumpfile.h (enum dump_kind): Likewise.
23335 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23336 * print-tree.c (debug_verbose): Remove.
23337 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23338 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23339 * tree-diagnostic.c (default_tree_printer): Replace
23340 TDF_DIAGNOSTIC with TDF_SLIM.
23342 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23344 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23345 in parameter loop, rather than loop_vinfo.
23346 (create_intersect_range_checks): Ditto.
23347 (vect_create_cond_for_alias_checks): Update call to above functions.
23349 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23351 PR tree-optimization/80815
23352 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23353 for merging runtime alias checks. Handle negative DR_STEPs.
23355 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23357 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23359 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23360 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23361 out code pruning runtime alias checks.
23362 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23363 factored out from above.
23364 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23366 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23368 (prune_runtime_alias_test_list): New decalaration.
23370 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23372 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23373 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23374 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23375 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23376 (operator==, comp_dr_with_seg_len_pair): Ditto.
23377 (vect_prune_runtime_alias_test_list): Ditto.
23379 2017-05-26 Martin Liska <mliska@suse.cz>
23382 * params.def: Bound partial-inlining-entry-probability param.
23384 2017-05-26 Marek Polacek <polacek@redhat.com>
23387 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23390 2017-05-26 Richard Biener <rguenther@suse.de>
23392 PR tree-optimization/80842
23393 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23396 2017-05-26 Richard Biener <rguenther@suse.de>
23398 PR tree-optimization/80844
23399 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23401 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23403 * doc/md.texi (Machine Constraints): Update x86 family
23404 machine constraints section to match 'config/i386/constraints.md'.
23406 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23408 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23410 2017-05-25 Nathan Sidwell <nathan@acm.org>
23412 * doc/invoke.texi (--enable-languages): Update documentation.
23414 2017-05-25 Martin Liska <mliska@suse.cz>
23416 * dumpfile.c: Add TDF_FOLDING.
23417 * dumpfile.h (enum dump_kind): Likewise.
23418 * genmatch.c (dt_simplify::gen_1): Use it.
23420 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23422 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23424 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23426 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23427 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23429 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23431 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23432 * match.pd (X == C): Rewrite it here.
23433 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23434 with_certain_nonzero_bits2): New predicates.
23435 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23437 2017-05-24 Nathan Sidwell <nathan@acm.org>
23439 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23442 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23445 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23447 * config/powerpcspe: New port. Files are copied from the rs6000
23448 port, with "rs6000" in filenames replaced by "powerpcspe".
23450 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23452 PR rtl-optimization/80754
23453 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23455 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23457 * config/sparc/sparc.md (length): Return the correct value for -mflat
23458 sibcalls to match output_sibcall.
23460 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23464 * config/rs6000/rs6000.c (struct machine_function): Add new field
23466 (rs6000_get_separate_components): Init that field, use it.
23467 (rs6000_components_for_bb): Use the field.
23469 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23471 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23473 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23475 PR middle-end/80823
23476 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23478 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23481 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23482 address against address_operand predicate.
23483 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23485 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23487 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23488 parameters passed indirectly.
23490 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23492 * config/i386/i386.md (*movdi_internal): Remove SSE4
23493 alternative 18 (?r, *v). Update insn attributes.
23494 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23495 Update insn attributes.
23496 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23497 Update insn attributes.
23498 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23499 alternative 1 (r, v). Remove isa attribute.
23500 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23501 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23502 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23504 2017-05-23 Tom de Vries <tom@codesourcery.com>
23506 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23509 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23511 * cgraphunit.c (symbol_table::process_new_functions): Update.
23512 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23513 (inline_generate_summary): Rename to ...
23514 (ipa_fn_summary_generate): ... this one.
23515 (inline_read_summary): Rename to ...
23516 (ipa_fn_summary_read): ... this one.
23517 (inline_write_summary): Rename to ...
23518 (ipa_fn_summary_write): ... this one.
23519 (inline_free_summary): Rename to ...
23520 (ipa_free_fn_summary): ... this one.
23521 (pass_data_local_fn_summary, pass_local_fn_summary,
23522 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23523 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23524 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23525 make_pass_ipa_fn_summary): New.
23526 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23527 inline_write_summary, inline_free_summary): Remove.
23528 (ipa_free_fn_summary) : New.
23529 * ipa-inline.c (ipa_inline): Update.
23530 (pass_ipa_inline): Do not generate summaries.
23531 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23533 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23534 and add pass_ipa_fn_summary.
23535 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23537 (make_pass_inline_parameters): Remove.
23539 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23541 * omp-low.c (struct omp_context): Remove "default_kind" member.
23544 * omp-offload.c (execute_oacc_device_lower): Remove the
23545 parallelism dimensions function attributes for unparallelized
23546 OpenACC kernels constructs.
23548 2017-05-23 Martin Liska <mliska@suse.cz>
23550 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23552 (cgraph_edge::make_speculative): Likewise.
23553 (cgraph_edge::resolve_speculation): Likewise.
23554 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23555 (cgraph_node::dump): Likewise.
23556 * cgraph.h: Likewise.
23557 * cgraphunit.c (analyze_functions): Likewise.
23558 (symbol_table::compile): Likewise.
23559 * ipa-cp.c (print_all_lattices): Likewise.
23560 (determine_versionability): Likewise.
23561 (initialize_node_lattices): Likewise.
23562 (ipcp_verify_propagated_values): Likewise.
23563 (estimate_local_effects): Likewise.
23564 (update_profiling_info): Likewise.
23565 (create_specialized_node): Likewise.
23566 (perhaps_add_new_callers): Likewise.
23567 (decide_about_value): Likewise.
23568 (decide_whether_version_node): Likewise.
23569 (identify_dead_nodes): Likewise.
23570 (ipcp_store_bits_results): Likewise.
23571 * ipa-devirt.c (dump_targets): Likewise.
23572 (ipa_devirt): Likewise.
23573 * ipa-icf.c (sem_item::dump): Likewise.
23574 (sem_function::equals): Likewise.
23575 (sem_variable::equals): Likewise.
23576 (sem_item_optimizer::read_section): Likewise.
23577 (sem_item_optimizer::execute): Likewise.
23578 (congruence_class::dump): Likewise.
23579 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23580 (dump_inline_summary): Likewise.
23581 (estimate_node_size_and_time): Likewise.
23582 (inline_analyze_function): Likewise.
23583 * ipa-inline-transform.c (inline_call): Likewise.
23584 * ipa-inline.c (report_inline_failed_reason): Likewise.
23585 (want_early_inline_function_p): Likewise.
23586 (edge_badness): Likewise.
23587 (update_edge_key): Likewise.
23588 (inline_small_functions): Likewise.
23589 * ipa-profile.c (ipa_profile): Likewise.
23590 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23591 (ipa_make_edge_direct_to_target): Likewise.
23592 (remove_described_reference): Likewise.
23593 (ipa_impossible_devirt_target): Likewise.
23594 (propagate_controlled_uses): Likewise.
23595 (ipa_print_node_params): Likewise.
23596 (ipcp_transform_function): Likewise.
23597 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23598 (propagate_pure_const): Likewise.
23599 * ipa-reference.c (generate_summary): Likewise.
23600 (read_write_all_from_decl): Likewise.
23601 (propagate): Likewise.
23602 (ipa_reference_read_optimization_summary): Likewise.
23603 * ipa-utils.c (ipa_merge_profiles): Likewise.
23604 * ipa.c (walk_polymorphic_call_targets): Likewise.
23605 (symbol_table::remove_unreachable_nodes): Likewise.
23606 (ipa_single_use): Likewise.
23607 * passes.c (execute_todo): Likewise.
23608 * predict.c (drop_profile): Likewise.
23609 * symtab.c (symtab_node::get_dump_name): New function.
23610 (symtab_node::dump_name): Likewise.
23611 (symtab_node::dump_asm_name): Likewise.
23612 (symtab_node::dump_references): Likewise.
23613 (symtab_node::dump_referring): Likewise.
23614 (symtab_node::dump_base): Likewise.
23615 (symtab_node::debug_symtab): Likewise.
23616 * tree-sra.c (convert_callers_for_node): Likewise.
23617 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23618 * value-prof.c (init_node_map): Likewise.
23620 2017-05-23 Martin Liska <mliska@suse.cz>
23622 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23623 and symtab_node::debug_symtab to symbol_table::debug.
23624 * cgraphunit.c (analyze_functions): Use the renamed function.
23625 (symbol_table::compile): Likewise.
23626 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23627 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23628 * passes.c (execute_todo): Likewise.
23629 * symtab.c (symbol_table::dump): New function.
23630 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23632 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23634 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23635 that nonconst implies exec.
23637 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23639 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23640 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23641 (inline_edge_summary_vec): Turn into ...
23642 (ipa_call_summaries): ... this one.
23643 (redirect_to_unreachable, edge_set_predicate,
23644 evaluate_properties_for_edge, inline_summary_alloc,
23645 reset_ipa_call_summary, reset_inline_summary,
23646 inline_summary_t::duplicate): Update.
23647 (inline_edge_duplication_hook): Turn to ...
23648 (ipa_call_summary_t::duplicate): ... this one.
23649 (inline_edge_removal_hook): Turn to ...
23650 (ipa_call_summary_t::remove): ... this one.
23651 (dump_inline_edge_summary): Turn to ...
23652 (dump_ipa_call_summary): ... this one.
23653 (estimate_function_body_sizes): Update.
23654 (inline_update_callee_summaries): Update.
23655 (remap_edge_change_prob): Update.
23656 (remap_edge_summaries): Update.
23657 (inline_merge_summary): Update.
23658 (do_estimate_edge_time): Update.
23659 (inline_generate_summary): Update.
23660 (inline_read_section): Update.
23661 (inline_read_summary): Update.
23662 (inline_free_summary): Update.
23663 * ipa-inline.c (can_inline_edge_p): Update.
23664 (compute_inlined_call_time): Update.
23665 (want_inline_small_function_p): Update.
23666 (edge_badness): Update.
23667 (early_inliner): Update.
23668 * ipa-inline.h (inline_edge_summary): Turn to ...
23669 (ipa_call_summary): ... this one.
23670 (ipa_call_summary_t): New class.
23671 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23672 (ipa_call_summaries): New.
23673 (inline_edge_summary): Remove.
23674 (estimate_edge_growth): Update.
23675 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23676 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23677 * ipa-split.c (execute_split_functions): Update.
23678 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23680 2017-05-23 Tom de Vries <tom@codesourcery.com>
23682 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23683 attributes): Document rdrand effective target.
23685 2017-05-23 Tom de Vries <tom@codesourcery.com>
23687 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23688 attributes): Sort alphabetically.
23690 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23692 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23694 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23697 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23698 V2DF/V2DI splat into two separate patterns, one that handles
23699 registers, and the other that only handles memory. Drop support
23700 for splatting from a GPR on ISA 2.07 and then splitting the
23701 splat into direct move and splat.
23702 (vsx_splat_<mode>_reg): Likewise.
23703 (vsx_splat_<mode>_mem): Likewise.
23705 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23707 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23709 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23711 PR middle-end/80809
23712 * omp-low.c (finish_taskreg_remap): New function.
23713 (finish_taskreg_scan): If unit size of ctx->record_type
23714 is non-constant, unshare the size expression and replace
23715 decls in it with possible outer var refs.
23717 PR middle-end/80809
23718 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23719 GOVD_SHARED rather than GOVD_PRIVATE with it.
23720 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23721 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23723 PR middle-end/80853
23724 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23725 as last argument to build_outer_var_ref for pointer bases of array
23726 section reductions.
23728 2017-05-19 Martin Sebor <msebor@redhat.com>
23730 * print-tree.c (print_node): Print DECL_READ_P flag.
23732 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23734 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23735 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23736 * cgraph.c: Likewise.
23737 * cgraphunit.c: Likewise.
23738 * gengtype.c: Likewise.
23739 * ipa-cp.c: Likewise.
23740 * ipa-devirt.c: Likewise.
23741 * ipa-icf.c: Likewise.
23742 * ipa-predicate.c: Likewise.
23743 * ipa-profile.c: Likewise.
23744 * ipa-prop.c: Likewise.
23745 * ipa-split.c: Likewise.
23747 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23748 edge_predicate_pool, dump_inline_hints,
23749 inline_summary::account_size_time, redirect_to_unreachable,
23750 edge_set_predicate, set_hint_predicate,
23751 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23752 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23753 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23754 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23755 ipa_call_summary_t::remove, initialize_growth_caches,
23756 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23757 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23758 mark_modified, unmodified_parm_1, unmodified_parm,
23759 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23760 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23761 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23762 will_be_nonconstant_predicate, record_modified_bb_info,
23763 get_minimal_bb, record_modified, param_change_prob,
23764 phi_result_unknown_predicate, predicate_for_phi_result,
23765 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23766 estimate_function_body_sizes, compute_inline_parameters,
23767 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23768 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23769 inline_update_callee_summaries, remap_edge_change_prob,
23770 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23771 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23772 inline_analyze_function, inline_summary_t::insert,
23773 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23774 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23775 inline_free_summary): Move to ipa-fnsummary.h
23776 (predicate_t): Remove.
23777 * ipa-fnsummary.c: New file.
23778 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23779 (enum inline_hints_vals, inline_hints, agg_position_info,
23780 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23781 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23782 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23783 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23784 inline_read_summary, inline_write_summary, inline_free_summary,
23785 inline_analyze_function, initialize_inline_failed,
23786 inline_merge_summary, inline_update_overall_summary,
23787 compute_inline_parameters): Move to ipa-fnsummary.h
23788 * ipa-fnsummary.h: New file.
23789 * ipa-inline-transform.h: Include ipa-inline.h.
23790 * ipa-inline.c: LIkewise.
23792 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23794 * ipa-inline.c (edge_badness): Use inlined_time instead of
23795 inline_summaries->get.
23797 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23799 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23801 2017-05-22 Nathan Sidwell <nathan@acm.org>
23803 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23804 (fdump-lang): Document 'raw' option.
23805 * dumpfile.h (TDI_tu): Delete.
23806 * dumpfile.c (dump_files): Remove translation-unit.
23807 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23809 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23811 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23812 command option from $(AWK) call.
23813 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23814 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23815 [FORMAT]: Remove handling of variable.
23816 * config/avr/t-multilib: Regenerate.
23818 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23820 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23822 (dump_inline_summary): Do not print self_time.
23823 (estimate_function_body_sizes): Do not set self_time.
23824 (compute_inline_parameters): Likewise.
23825 (inline_read_section, inline_write_summary): Do not stream self_time.
23826 * ipa-inline.h (inline_summary): Drop self_time.
23828 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23830 * ipa-inline-analysis.c (account_size_time): Rename to ...
23831 (inline_summary::account_size_time): ... this one.
23832 (reset_ipa_call_summary): Turn to ...
23833 (ipa_call_summary::reset): ... this one.
23834 (reset_inline_summary): Turn to ...
23835 (inline_summary::reset): ... this one.
23836 (inline_summary_t::remove): Update.
23837 (inline_summary_t::duplicate): Update.
23838 (ipa_call_summary_t::remove): Update.
23839 (dump_inline_summary): Update.
23840 (estimate_function_body_sizes): Update.
23841 (compute_inline_parameters): Update.
23842 (estimate_node_size_and_time): Update.
23843 (inline_merge_summary): Update.
23844 (inline_update_overall_summary): Update.
23845 (inline_read_section): Update.
23846 (inline_write_summary): Update.
23847 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23848 add account_size_time and reset member functions.
23849 (ipa_call_summary): Add reset function.
23850 * ipa-predicate.h (predicate::operator &): Constify.
23852 2017-05-22 Richard Biener <rguenther@suse.de>
23854 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23856 2017-05-19 Jason Merrill <jason@redhat.com>
23858 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23860 2017-05-19 Marek Polacek <polacek@redhat.com>
23863 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23864 TYPE_OVERFLOW_WRAPS checks.
23866 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23868 * tree-core.h (enum omp_clause_default_kind): Add
23869 "OMP_CLAUSE_DEFAULT_PRESENT".
23870 * tree-pretty-print.c (dump_omp_clause): Handle it.
23871 * gimplify.c (enum gimplify_omp_var_data): Add
23872 "GOVD_MAP_FORCE_PRESENT".
23873 (gimplify_adjust_omp_clauses_1): Map it to
23874 "GOMP_MAP_FORCE_PRESENT".
23875 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23877 * gimplify.c (oacc_default_clause): Clarify.
23879 2017-05-19 Nathan Sidwell <nathan@acm.org>
23881 LANG_HOOK_REGISTER_DUMPS
23882 * toplev.c (general_init): Call register dump lang hook.
23883 * doc/invoke.texi: Document -fdump-lang option family.
23884 * dumpfile.c (dump_files): Remove class dump here.
23885 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23886 * dumpfile.h (tree_dump_index): Remove TDI_class.
23887 * langhooks-def.h (lhd_register_dumps): Declare.
23888 (LANG_HOOKS_REGISTER_DUMPS): Define.
23889 (LANG_HOOKS_INITIALIZER): Add it.
23890 * langhooks.c (lhd_register_dumps): Define.
23891 * langhooks.h (struct lang_hooks): Add register_dumps.
23893 2017-05-19 Nathan Sidwell <nathan@acm.org>
23895 * context.h (context::set_passes): New.
23896 * context.c (context::context): Do not create pass manager.
23897 * toplev.c (general_init): Create pass manager here.
23899 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23901 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23902 use this splitter if two add or or instructions would also work for
23903 the constant we want to generate.
23905 2017-05-19 Richard Biener <rguenther@suse.de>
23908 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23909 predicate evaluation.
23911 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23913 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23915 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23918 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23920 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23921 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23922 (inline_edge_summary_vec): Turn into ...
23923 (ipa_call_summaries): ... this one.
23924 (redirect_to_unreachable, edge_set_predicate,
23925 evaluate_properties_for_edge, inline_summary_alloc,
23926 reset_ipa_call_summary, reset_inline_summary,
23927 inline_summary_t::duplicate): Update.
23928 (inline_edge_duplication_hook): Turn to ...
23929 (ipa_call_summary_t::duplicate): ... this one.
23930 (inline_edge_removal_hook): Turn to ...
23931 (ipa_call_summary_t::remove): ... this one.
23932 (dump_inline_edge_summary): Turn to ...
23933 (dump_ipa_call_summary): ... this one.
23934 (estimate_function_body_sizes): Update.
23935 (inline_update_callee_summaries): Update.
23936 (remap_edge_change_prob): Update.
23937 (remap_edge_summaries): Update.
23938 (inline_merge_summary): Update.
23939 (do_estimate_edge_time): Update.
23940 (inline_generate_summary): Update.
23941 (inline_read_section): Update.
23942 (inline_read_summary): Update.
23943 (inline_free_summary): Update.
23944 * ipa-inline.c (can_inline_edge_p): Update.
23945 (compute_inlined_call_time): Update.
23946 (want_inline_small_function_p): Update.
23947 (edge_badness): Update.
23948 (early_inliner): Update.
23949 * ipa-inline.h (inline_edge_summary): Turn to ...
23950 (ipa_call_summary): ... this one.
23951 (ipa_call_summary_t): New class.
23952 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23953 (ipa_call_summaries): New.
23954 (inline_edge_summary): Remove.
23955 (estimate_edge_growth): Update.
23956 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23957 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23958 * ipa-split.c (execute_split_functions): Update.
23959 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23961 2017-05-19 Richard Biener <rguenther@suse.de>
23963 PR middle-end/80764
23964 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23966 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23968 * config/rs6000/rs6000.c (struct machine_function): Add field
23969 fpr_is_wrapped_separately.
23970 (rs6000_get_separate_components): Use 64 components. Handle the
23971 new FPR components.
23972 (rs6000_components_for_bb): Handle the FPR components.
23973 (rs6000_emit_prologue_components): Handle the FPR components.
23974 (rs6000_emit_epilogue_components): Handle the FPR components.
23975 (rs6000_set_handled_components): Handle the FPR components.
23976 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23977 that are already separately shrink-wrapped.
23978 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23979 that are already separately shrink-wrapped.
23981 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23984 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23987 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23988 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23989 cases where the register allocator uses a move and an offsettable
23990 memory operation to/from a FPR register on ISA 2.06/2.07.
23991 (define_peephole2 for Altivec d-form store): Likewise.
23993 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23996 * config/i386/mmx.md (*mov<mode>_internal): Enable
23997 alternatives 11, 12, 13 and 14 also for 32bit targets.
23998 Remove alternatives 15, 16, 17 and 18.
23999 * config/i386/sse.md (vec_concatv2di): Change
24000 alternative (!x, *y) to (x, ?!*Yn).
24002 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24004 * dumpfile.h (enum dump_kind): Remove stray comma.
24006 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24008 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24009 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24010 predicate::num_conditions
24011 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24012 (CHANGED): turn into predicate::changed.
24013 (agg_position_info): Move to ipa-predicate.h
24014 (add_condition, predicate::add_clause, predicate::operator &=,
24015 predicate::or_with, predicate::evaluate, predicate::probability,
24016 dump_condition, dump_clause, predicate::dump,
24017 predicate::remap_after_duplication, predicate::remap_after_inlining,
24018 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24019 (evaluate_conditions_for_known_args): Update.
24020 (set_cond_stmt_execution_predicate): Update.
24021 * ipa-inline.h: Include ipa-predicate.h
24022 (condition, inline_param_summary, conditions, agg_position_info,
24023 predicate): Move to ipa-predicate.h
24024 * ipa-predicate.c: New file.
24025 * ipa-predicate.h: New file.
24027 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24029 * final.c (leaf_function_p): Check we are not in a sequence.
24031 2017-05-18 Martin Liska <mliska@suse.cz>
24033 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24034 * dumpfile.c (dump_register): Use new enum dump_kind.
24035 (get_dump_file_name): Likewise.
24036 (dump_enable_all): Likewise.
24037 (dump_switch_p_1): Likewise.
24038 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24039 * dumpfile.h (enum dump_kind): New enum type.
24040 (struct dump_file_info): Create constructor and
24041 format fields and comments.
24042 * passes.c (pass_manager::register_one_dump_file):
24044 * statistics.c (statistics_early_init): Likewise.
24045 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24046 TDF_TREE with TDF_SLIM.
24047 (gather_memory_references_ref): Likewise.
24049 2017-05-18 Martin Liska <mliska@suse.cz>
24051 * vec.h (struct vnull): Use it.
24053 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24055 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24056 (true_predicate, false_predicate, true_predicate_p,
24057 false_predicate_p): Remove.
24058 (single_cond_predicate, not_inlined_predicate): Turn to member function
24060 (add_condition): Update.
24061 (add_clause): Turn to...
24062 (predicate::add_clause): ... this one; update; allow passing NULL
24064 (and_predicates): Turn to ...
24065 (predicate::operator &=): ... this one.
24066 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24067 (or_predicates): Turn to ...
24068 (predicate::or_with): ... this one.
24069 (evaluate_predicate): Turn to ...
24070 (predicate::evaluate): ... this one.
24071 (predicate_probability): Turn to ...
24072 (predicate::probability): ... this one.
24073 (dump_condition): Update.
24074 (dump_predicate): Turn to ...
24075 (predicate::dump): ... this one.
24076 (account_size_time): Update.
24077 (edge_set_predicate): Update.
24078 (set_hint_predicate): UPdate.
24079 (evaluate_conditions_for_known_args): Update.
24080 (evaluate_properties_for_edge): Update.
24081 (remap_predicate_after_duplication): Turn to...
24082 (predicate::remap_after_duplication): ... this one.
24083 (remap_hint_predicate_after_duplication): Update.
24084 (inline_summary_t::duplicate): UPdate.
24085 (dump_inline_edge_summary): Update.
24086 (dump_inline_summary): Update.
24087 (set_cond_stmt_execution_predicate): Update.
24088 (set_switch_stmt_execution_predicate): Update.
24089 (compute_bb_predicates): Update.
24090 (will_be_nonconstant_expr_predicate): Update.
24091 (will_be_nonconstant_predicate): Update.
24092 (phi_result_unknown_predicate): Update.
24093 (predicate_for_phi_result): Update.
24094 (array_index_predicate): Update.
24095 (estimate_function_body_sizes): Update.
24096 (estimate_node_size_and_time): Update.
24097 (estimate_ipcp_clone_size_and_time): Update.
24098 (remap_predicate): Rename to ...
24099 (predicate::remap_after_inlining): ... this one.
24100 (remap_hint_predicate): Update.
24101 (inline_merge_summary): Update.
24102 (inline_update_overall_summary): Update.
24103 (estimate_size_after_inlining): Update.
24104 (read_predicate): Rename to ...
24105 (predicate::stream_in): ... this one.
24106 (read_inline_edge_summary): Update.
24107 (write_predicate): Rename to ...
24108 (predicate::stream_out): ... this one.
24109 (write_inline_edge_summary): Update.
24110 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24111 (clause_t): Turn to uint32_t
24112 (predicate): Turn to class; implement constructor and operators
24114 (size_time_entry): Update.
24115 (inline_summary): Update.
24116 (inline_edge_summary): Update.
24118 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24120 * fold-const.c (fold_binary_loc): Move transformation...
24121 * match.pd (C - X CMP X): ... here.
24123 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24125 * config/sparc/sparc.c (sparc_option_override): Set function
24126 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24127 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24129 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24131 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24133 2017-05-18 Marek Polacek <polacek@redhat.com>
24136 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24137 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24138 gimple_assign_load_p.
24140 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24142 PR middle-end/80692
24143 * real.c (do_compare): Give decimal_do_compare preference over
24144 comparing just the signs.
24146 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24148 * doc/md.texi (Canonicalization of Instructions): Describe the
24149 canonical form of instructions that inherently set a condition
24152 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24154 PR middle-end/80775
24155 * tree-cfg.c: Move deletion of unreachable case statements to after
24156 the merging of consecutive case labels.
24158 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24160 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24161 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24162 restoring of callee-saved registers.
24164 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24166 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24167 * config/visium/visium.c (single_set_and_flags): Likewise.
24168 * config/visium/visium.md (Substitutions): Likewise.
24170 2017-05-17 Martin Liska <mliska@suse.cz>
24172 * cfg.c: Introduce dump_flags_t type and
24173 use it instead of int type.
24175 * cfghooks.c: Likewise.
24176 * cfghooks.h (struct cfg_hooks): Likewise.
24177 * cfgrtl.c: Likewise.
24178 * cfgrtl.h: Likewise.
24179 * cgraph.c (cgraph_node::get_body): Likewise.
24180 * coretypes.h: Likewise.
24181 * domwalk.c: Likewise.
24182 * domwalk.h: Likewise.
24183 * dumpfile.c (struct dump_option_value_info): Likewise.
24184 (dump_enable_all): Likewise.
24185 (dump_switch_p_1): Likewise.
24186 (opt_info_switch_p): Likewise.
24187 * dumpfile.h (enum tree_dump_index): Likewise.
24188 (struct dump_file_info): Likewise.
24189 * genemit.c: Likewise.
24190 * generic-match-head.c: Likewise.
24191 * gengtype.c (open_base_files): Likewise.
24192 * gimple-pretty-print.c: Likewise.
24193 * gimple-pretty-print.h: Likewise.
24194 * graph.c (print_graph_cfg): Likewise.
24195 * graphite-scop-detection.c (dot_all_sese): Likewise.
24196 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24197 * loop-unroll.c (report_unroll): Likewise.
24198 * passes.c (pass_manager::register_one_dump_file): Likewise.
24199 * print-tree.c: Likewise.
24200 * statistics.c: Likewise.
24201 * tree-cfg.c: Likewise.
24202 * tree-cfg.h: Likewise.
24203 * tree-dfa.c: Likewise.
24204 * tree-dfa.h: Likewise.
24205 * tree-dump.c (dump_function): Likewise.
24206 * tree-dump.h (struct dump_info): Likewise.
24207 * tree-pretty-print.c: Likewise.
24208 * tree-pretty-print.h: Likewise.
24209 * tree-ssa-live.c: Likewise.
24210 * tree-ssa-live.h: Likewise.
24211 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24212 * tree-vect-loop.c: Likewise.
24213 * tree-vect-slp.c: Likewise.
24215 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24216 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24218 PR tree-optimization/80457
24219 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24220 of all arguments to a statement as scalar_to_vec operations.
24221 (vectorizable_call): Adjust call to vect_model_simple_cost for
24223 (vectorizable_conversion): Likewise.
24224 (vectorizable_assignment): Likewise.
24225 (vectorizable_shift): Likewise.
24226 (vectorizable_operation): Likewise.
24227 (vectorizable_comparison): Likewise.
24228 (vect_is_simple_cond): Record the def types for operands.
24229 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24230 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24231 for statement argument count.
24233 2017-05-16 Carl Love <cel@us.ibm.com>
24235 * config/rs6000/rs6000-c: Add support for built-in functions
24236 vector unsigned long long vec_bperm (vector unsigned long long,
24237 vector unsigned char)
24238 vector signed long long vec_mule (vector signed int,
24240 vector unsigned long long vec_mule (vector unsigned int,
24241 vector unsigned int)
24242 vector signed long long vec_mulo (vector signed int,
24244 vector unsigned long long vec_mulo (vector unsigned int,
24245 vector unsigned int)
24246 vector signed char vec_sldw (vector signed char,
24247 vector signed char,
24249 vector unsigned char vec_sldw (vector unsigned char,
24250 vector unsigned char,
24252 vector signed short vec_sldw (vector signed short,
24253 vector signed short,
24255 vector unsigned short vec_sldw (vector unsigned short,
24256 vector unsigned short,
24258 vector signed int vec_sldw (vector signed int,
24261 vector unsigned int vec_sldw (vector unsigned int,
24262 vector unsigned int,
24264 vector signed long long vec_sldw (vector signed long long,
24265 vector signed long long,
24267 vector unsigned long long vec_sldw (vector unsigned long long,
24268 vector unsigned long long,
24270 * config/rs6000/rs6000-c: Add support for built-in functions
24271 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24272 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24273 * doc/extend.texi: Update the built-in documentation for the
24274 new built-in functions.
24276 2017-05-16 Marek Polacek <polacek@redhat.com>
24280 * tree.c (save_expr): Don't fold the expression.
24282 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24284 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24285 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24286 and (?*y,m). Update insn attributes.
24288 2017-05-16 Martin Liska <mliska@suse.cz>
24290 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24291 flags argument of print_gimple_stmt, print_gimple_expr,
24292 print_generic_stmt and print_generic_expr.
24293 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24294 * coretypes.h: Likewise.
24295 * except.c (dump_eh_tree): Likewise.
24296 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24297 * gimple-pretty-print.h: Likewise.
24298 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24299 (backprop::push_to_worklist): Likewise.
24300 (backprop::pop_from_worklist): Likewise.
24301 (backprop::process_use): Likewise.
24302 (backprop::intersect_uses): Likewise.
24303 (note_replacement): Likewise.
24304 * gimple-ssa-store-merging.c
24305 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24306 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24307 (pass_store_merging::execute): Likewise.
24308 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24309 (ssa_base_cand_dump_callback): Likewise.
24310 (dump_incr_vec): Likewise.
24311 (replace_refs): Likewise.
24312 (replace_mult_candidate): Likewise.
24313 (create_add_on_incoming_edge): Likewise.
24314 (create_phi_basis): Likewise.
24315 (insert_initializers): Likewise.
24316 (all_phi_incrs_profitable): Likewise.
24317 (introduce_cast_before_cand): Likewise.
24318 (replace_one_candidate): Likewise.
24319 * gimplify.c (gimplify_expr): Likewise.
24320 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24321 (set_rename): Likewise.
24322 (rename_uses): Likewise.
24323 (copy_loop_phi_nodes): Likewise.
24324 (add_close_phis_to_merge_points): Likewise.
24325 (copy_loop_close_phi_args): Likewise.
24326 (copy_cond_phi_args): Likewise.
24327 (graphite_copy_stmts_from_block): Likewise.
24328 (translate_pending_phi_nodes): Likewise.
24329 * graphite-poly.c (print_pdr): Likewise.
24330 (dump_gbb_cases): Likewise.
24331 (dump_gbb_conditions): Likewise.
24332 (print_scop_params): Likewise.
24333 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24334 (build_cross_bb_scalars_use): Likewise.
24335 (gather_bbs::before_dom_children): Likewise.
24336 * hsa-dump.c (dump_hsa_immed): Likewise.
24337 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24338 (get_replacement_map): Likewise.
24339 * ipa-inline-analysis.c (dump_condition): Likewise.
24340 (estimate_function_body_sizes): Likewise.
24341 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24342 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24343 * ipa-prop.c (ipa_dump_param): Likewise.
24344 (ipa_print_node_jump_functions_for_edge): Likewise.
24345 (ipa_modify_call_arguments): Likewise.
24346 (ipa_modify_expr): Likewise.
24347 (ipa_dump_param_adjustments): Likewise.
24348 (ipa_dump_agg_replacement_values): Likewise.
24349 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24350 * ipa-pure-const.c (check_stmt): Likewise.
24351 (pass_nothrow::execute): Likewise.
24352 * ipa-split.c (execute_split_functions): Likewise.
24353 * omp-offload.c (dump_oacc_loop_part): Likewise.
24354 (dump_oacc_loop): Likewise.
24355 * trans-mem.c (tm_log_emit): Likewise.
24356 (tm_memopt_accumulate_memops): Likewise.
24357 (dump_tm_memopt_set): Likewise.
24358 (dump_tm_memopt_transform): Likewise.
24359 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24360 (print_loop): Likewise.
24361 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24362 (chkp_gather_checks_info): Likewise.
24363 (chkp_get_check_result): Likewise.
24364 (chkp_remove_check_if_pass): Likewise.
24365 (chkp_use_outer_bounds_if_possible): Likewise.
24366 (chkp_reduce_bounds_lifetime): Likewise.
24367 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24368 (chkp_mark_completed_bounds): Likewise.
24369 (chkp_register_incomplete_bounds): Likewise.
24370 (chkp_mark_invalid_bounds): Likewise.
24371 (chkp_maybe_copy_and_register_bounds): Likewise.
24372 (chkp_build_returned_bound): Likewise.
24373 (chkp_get_bound_for_parm): Likewise.
24374 (chkp_build_bndldx): Likewise.
24375 (chkp_get_bounds_by_definition): Likewise.
24376 (chkp_generate_extern_var_bounds): Likewise.
24377 (chkp_get_bounds_for_decl_addr): Likewise.
24378 * tree-chrec.c (chrec_apply): Likewise.
24379 * tree-data-ref.c (dump_data_reference): Likewise.
24380 (dump_subscript): Likewise.
24381 (dump_data_dependence_relation): Likewise.
24382 (analyze_overlapping_iterations): Likewise.
24383 * tree-inline.c (expand_call_inline): Likewise.
24384 (tree_function_versioning): Likewise.
24385 * tree-into-ssa.c (dump_defs_stack): Likewise.
24386 (dump_currdefs): Likewise.
24387 (dump_names_replaced_by): Likewise.
24388 (dump_update_ssa): Likewise.
24389 (update_ssa): Likewise.
24390 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24391 * tree-parloops.c (build_new_reduction): Likewise.
24392 (try_create_reduction_list): Likewise.
24393 (ref_conflicts_with_region): Likewise.
24394 (oacc_entry_exit_ok_1): Likewise.
24395 (oacc_entry_exit_single_gang): Likewise.
24396 * tree-pretty-print.h: Likewise.
24397 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24398 (get_scalar_evolution): Likewise.
24399 (add_to_evolution): Likewise.
24400 (get_loop_exit_condition): Likewise.
24401 (analyze_evolution_in_loop): Likewise.
24402 (analyze_initial_condition): Likewise.
24403 (analyze_scalar_evolution): Likewise.
24404 (instantiate_scev): Likewise.
24405 (number_of_latch_executions): Likewise.
24406 (gather_chrec_stats): Likewise.
24407 (final_value_replacement_loop): Likewise.
24408 (scev_const_prop): Likewise.
24409 * tree-sra.c (dump_access): Likewise.
24410 (disqualify_candidate): Likewise.
24411 (create_access): Likewise.
24412 (reject): Likewise.
24413 (maybe_add_sra_candidate): Likewise.
24414 (create_access_replacement): Likewise.
24415 (analyze_access_subtree): Likewise.
24416 (analyze_all_variable_accesses): Likewise.
24417 (sra_modify_assign): Likewise.
24418 (initialize_constant_pool_replacements): Likewise.
24419 (find_param_candidates): Likewise.
24420 (decide_one_param_reduction): Likewise.
24421 (replace_removed_params_ssa_names): Likewise.
24422 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24423 * tree-ssa-copy.c (dump_copy_of): Likewise.
24424 (copy_prop_visit_cond_stmt): Likewise.
24425 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24426 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24427 (record_equivalences_from_stmt): Likewise.
24428 * tree-ssa-dse.c (compute_trims): Likewise.
24429 (delete_dead_call): Likewise.
24430 (delete_dead_assignment): Likewise.
24431 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24432 (forward_propagate_into_cond): Likewise.
24433 (pass_forwprop::execute): Likewise.
24434 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24435 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24437 (move_computations_worker): Likewise.
24438 (execute_sm): Likewise.
24439 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24440 (remove_exits_and_undefined_stmts): Likewise.
24441 (remove_redundant_iv_tests): Likewise.
24442 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24443 (adjust_iv_update_pos): Likewise.
24444 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24445 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24446 (value_replacement): Likewise.
24447 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24448 * tree-ssa-pre.c (print_pre_expr): Likewise.
24449 (get_representative_for): Likewise.
24450 (create_expression_by_pieces): Likewise.
24451 (insert_into_preds_of_block): Likewise.
24452 (eliminate_insert): Likewise.
24453 (eliminate_dom_walker::before_dom_children): Likewise.
24454 (eliminate): Likewise.
24455 (remove_dead_inserted_code): Likewise.
24456 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24457 * tree-ssa-reassoc.c (get_rank): Likewise.
24458 (eliminate_duplicate_pair): Likewise.
24459 (eliminate_plus_minus_pair): Likewise.
24460 (eliminate_not_pairs): Likewise.
24461 (undistribute_ops_list): Likewise.
24462 (eliminate_redundant_comparison): Likewise.
24463 (update_range_test): Likewise.
24464 (optimize_range_tests_var_bound): Likewise.
24465 (optimize_vec_cond_expr): Likewise.
24466 (rewrite_expr_tree): Likewise.
24467 (rewrite_expr_tree_parallel): Likewise.
24468 (linearize_expr): Likewise.
24469 (break_up_subtract): Likewise.
24470 (linearize_expr_tree): Likewise.
24471 (attempt_builtin_powi): Likewise.
24472 (attempt_builtin_copysign): Likewise.
24473 (transform_stmt_to_copy): Likewise.
24474 (transform_stmt_to_multiply): Likewise.
24475 (dump_ops_vector): Likewise.
24476 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24477 (print_scc): Likewise.
24478 (set_ssa_val_to): Likewise.
24479 (visit_reference_op_store): Likewise.
24480 (visit_use): Likewise.
24481 (sccvn_dom_walker::before_dom_children): Likewise.
24482 (run_scc_vn): Likewise.
24483 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24485 (expr_hash_elt::print): Likewise.
24486 (const_and_copies::pop_to_marker): Likewise.
24487 (const_and_copies::record_const_or_copy_raw): Likewise.
24488 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24489 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24490 (dump_predicates): Likewise.
24491 (find_uninit_use): Likewise.
24492 (warn_uninitialized_phi): Likewise.
24493 (pass_late_warn_uninitialized::execute): Likewise.
24494 * tree-ssa.c (verify_vssa): Likewise.
24495 (verify_ssa): Likewise.
24496 (maybe_optimize_var): Likewise.
24497 * tree-vrp.c (dump_value_range): Likewise.
24498 (dump_all_value_ranges): Likewise.
24499 (dump_asserts_for): Likewise.
24500 (register_edge_assert_for_2): Likewise.
24501 (vrp_visit_cond_stmt): Likewise.
24502 (vrp_visit_switch_stmt): Likewise.
24503 (vrp_visit_stmt): Likewise.
24504 (vrp_visit_phi_node): Likewise.
24505 (simplify_cond_using_ranges_1): Likewise.
24506 (fold_predicate_in): Likewise.
24507 (evrp_dom_walker::before_dom_children): Likewise.
24508 (evrp_dom_walker::push_value_range): Likewise.
24509 (evrp_dom_walker::pop_value_range): Likewise.
24510 (execute_early_vrp): Likewise.
24512 2017-05-16 Richard Biener <rguenther@suse.de>
24514 * dwarf2out.c (loc_list_from_tree_1): Do not create
24515 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24517 2017-05-16 Richard Biener <rguenther@suse.de>
24519 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24521 (note_variable_value_in_expr): If we resolved the decl ref
24522 do not push to the stack.
24524 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24526 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24527 operations in fast-math mode.
24528 (vaddq_f16): Likewise.
24529 (vmul_f16): Likewise.
24530 (vmulq_f16): Likewise.
24531 (vsub_f16): Likewise.
24532 (vsubq_f16): Likewise.
24533 * config/arm/neon.md (add<mode>3): New.
24535 (fma:<VH:mode>3): New. Also remove outdated comment.
24538 2017-05-16 Martin Liska <mliska@suse.cz>
24542 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24543 (odr_types_equivalent_p): Likewise.
24545 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24547 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24549 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24552 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24553 non-interunit SSE move alternatives with '?'.
24554 (zero-extendsidi peephole2): New peephole to skip intermediate
24555 general register in SSE zero-extend sequence.
24557 2017-05-15 Jeff Law <law@redhat.com>
24559 * reorg.c (relax_delay_slots): Create a new variable to hold
24560 the temporary target rather than clobbering TARGET_LABEL.
24562 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24563 missing argument to extract_bit_field call.
24564 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24566 2017-05-15 Martin Liska <mliska@suse.cz>
24569 * gcc.c (process_command): Do not allow empty argument of -o option.
24571 2017-05-15 Renlin Li <renlin.li@arm.com>
24573 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24574 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24575 * config/aarch64/constraints.md (Usf): Add long call check.
24576 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24577 (call_value): Likewise.
24578 (sibcall): Likewise.
24579 (sibcall_value): Likewise.
24581 (call_value_insn): New.
24582 (sibcall_insn): Update rtx pattern.
24583 (sibcall_value_insn): Likewise.
24584 (call_internal): Remove.
24585 (call_value_internal): Likewise.
24586 (sibcall_internal): Likewise.
24587 (sibcall_value_internal): Likewise.
24588 (call_reg): Likewise.
24589 (call_symbol): Likewise.
24590 (call_value_reg): Likewise.
24591 (call_value_symbol): Likewise.
24593 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24596 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24598 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24600 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24601 compatible with CCGOCmode and with CCZmode.
24603 2017-05-14 Martin Sebor <msebor@redhat.com>
24605 PR middle-end/77671
24606 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24607 (gimple_fold_builtin_snprintf): Same.
24608 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24609 (gimple_fold_builtin_snprintf): Same.
24610 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24611 of character types.
24612 (is_call_safe): New function.
24613 (try_substitute_return_value): Call it.
24614 (try_simplify_call): New function.
24615 (pass_sprintf_length::handle_gimple_call): Call it.
24617 2017-05-14 Martin Sebor <msebor@redhat.com>
24619 PR middle-end/80669
24620 * builtins.c (expand_builtin_stpncpy): Simplify.
24622 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24624 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24625 * config/i386/i386.h
24626 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24627 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24628 (struct machine_function): Add new members call_ms2sysv,
24629 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24630 (struct machine_frame_state): New fields sp_realigned and
24631 sp_realigned_offset.
24632 * config/i386/i386.c
24633 (enum xlogue_stub): New enum.
24634 (enum xlogue_stub_sets): New enum.
24635 (class xlogue_layout): New class.
24636 (struct ix86_frame): New fields stack_realign_allocate_offset,
24637 stack_realign_offset and outlined_save_offset. Modify comments to
24638 detail stack layout when using out-of-line stubs.
24639 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24640 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24641 -mcall-ms2sysv-xlogues.
24642 (stub_managed_regs): New static variable.
24643 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24644 registers managed by out-of-line stub.
24645 (disable_call_ms2sysv_xlogues): New function.
24646 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24647 m->call_ms2sysv when appropriate and compute frame layout for
24649 (sp_valid_at, fp_valid_at): New inline functions.
24650 (choose_basereg): New function.
24651 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24653 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24654 Use align parameter of choose_baseaddr to generated aligned SSE movs
24656 (pro_epilogue_adjust_stack): Modify to track
24657 machine_frame_state::sp_realigned.
24658 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24659 (ix86_nsaved_sseregs): Likewise.
24660 (ix86_emit_save_regs): Likewise.
24661 (ix86_emit_save_regs_using_mov): Likewise.
24662 (ix86_emit_save_sse_regs_using_mov): Likewise.
24663 (get_scratch_register_on_entry): Likewise.
24664 (gen_frame_set): New function.
24665 (gen_frame_load): Likewise.
24666 (gen_frame_store): Likewise.
24667 (emit_outlined_ms2sysv_save): Likewise.
24668 (emit_outlined_ms2sysv_restore): Likewise.
24669 (ix86_expand_prologue): Modify stack re-alignment code and call
24670 emit_outlined_ms2sysv_save when appropriate.
24671 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24672 parameter rtx_insn *insn, which allows the function to be used to only
24673 generate the notes.
24674 (ix86_expand_epilogue): Modify validity checks of frame and stack
24675 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24676 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24677 * config/i386/predicates.md
24678 (save_multiple): New predicate.
24679 (restore_multiple): Likewise.
24680 * config/i386/sse.md
24681 (save_multiple<mode>): New pattern.
24682 (save_multiple_realign<mode>): Likewise.
24683 (restore_multiple<mode>): Likewise.
24684 (restore_multiple_and_return<mode>): Likewise.
24685 (restore_multiple_leave_return<mode>): Likewise.
24686 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24688 2017-05-14 Julia Koval <julia.koval@intel.com>
24690 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24691 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24692 (__builtin_ia32_xsetbv): New builtins.
24693 * config/i386/i386.c (ix86_expand_special_args_builtin):
24695 (ix86_expand_builtin): Special expand for new intrinsics.
24696 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24697 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24698 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24700 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24702 * cfganal.c (inverted_post_order_compute): Change argument type
24704 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24705 * df-core.c (rest_of_handle_df_initialize): Adjust.
24706 (rest_of_handle_df_finish): Likewise.
24707 (df_analyze_1): Likewise.
24708 (df_analyze): Likewise.
24709 (loop_inverted_post_order_compute): Change argument to be a vec *.
24710 (df_analyze_loop): Adjust.
24711 (df_get_n_blocks): Likewise.
24712 (df_get_postorder): Likewise.
24713 * df.h (struct df_d): Change field to be a vec.
24714 * lcm.c (compute_laterin): Adjust.
24715 (compute_available): Likewise.
24716 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24717 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24718 * tree-ssa-pre.c (compute_antic): Likewise.
24720 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24722 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24723 (depth_first_search::depth_first_search): Change structure init
24724 function to this constructor.
24725 (depth_first_search::add_bb): Rename function to this member.
24726 (depth_first_search::execute): Likewise.
24727 (flow_dfs_compute_reverse_finish): Adjust.
24729 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24731 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24732 (longest_simple_path): Likewise.
24733 * shrink-wrap.c (spread_components): Likewise.
24734 (disqualify_problematic_components): Likewise.
24735 (emit_common_heads_for_components): Likewise.
24736 (emit_common_tails_for_components): Likewise.
24737 (insert_prologue_epilogue_for_components): Likewise.
24739 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24741 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24744 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24746 * df-core.c (df_set_blocks): Start using auto_bitmap.
24747 (df_compact_blocks): Likewise.
24748 * df-problems.c (df_rd_confluence_n): Likewise.
24749 * df-scan.c (df_insn_rescan_all): Likewise.
24750 (df_process_deferred_rescans): Likewise.
24751 (df_update_entry_block_defs): Likewise.
24752 (df_update_exit_block_uses): Likewise.
24753 (df_entry_block_bitmap_verify): Likewise.
24754 (df_exit_block_bitmap_verify): Likewise.
24755 (df_scan_verify): Likewise.
24756 * lra-constraints.c (lra_constraints): Likewise.
24757 (undo_optional_reloads): Likewise.
24758 (lra_undo_inheritance): Likewise.
24759 * lra-remat.c (calculate_gen_cands): Likewise.
24760 (do_remat): Likewise.
24761 * lra-spills.c (assign_spill_hard_regs): Likewise.
24762 (spill_pseudos): Likewise.
24763 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24764 (bitmap_set_subtract_values): Likewise.
24766 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24768 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24769 management with auto_bitmap.
24770 (fix_inter_tick): Likewise.
24771 (fix_recovery_deps): Likewise.
24772 * ira.c (add_store_equivs): Likewise.
24773 (find_moveable_pseudos): Likewise.
24774 (split_live_ranges_for_shrink_wrap): Likewise.
24775 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24776 (rtx_reuse_manager::seen_def_p): Likewise.
24777 (rtx_reuse_manager::set_seen_def): Likewise.
24778 * print-rtl.h (class rtx_reuse_manager): Likewise.
24780 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24782 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24784 (migrate_btr_def): Likewise.
24785 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24786 * df-core.c (loop_post_order_compute): Likewise.
24787 (loop_inverted_post_order_compute): Likewise.
24788 * hsa-common.h: Likewise.
24789 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24790 * init-regs.c (initialize_uninitialized_regs): Likewise.
24791 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24792 (inline_small_functions): Likewise.
24793 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24794 * ira.c (combine_and_move_insns): Likewise.
24795 (build_insn_chain): Likewise.
24796 * loop-invariant.c (find_invariants): Likewise.
24797 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24798 * predict.c (tree_predict_by_opcode): Likewise.
24799 (predict_paths_leading_to): Likewise.
24800 (predict_paths_leading_to_edge): Likewise.
24801 (estimate_loops_at_level): Likewise.
24802 (estimate_loops): Likewise.
24803 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24804 (spread_components): Likewise.
24805 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24806 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24807 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24808 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24809 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24810 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24811 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24812 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24813 (mark_threaded_blocks): Likewise.
24814 (thread_through_all_blocks): Likewise.
24815 * tree-ssa.c (verify_ssa): Likewise.
24816 (execute_update_addresses_taken): Likewise.
24817 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24819 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24821 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24823 (post_order_compute): Likewise.
24824 (inverted_post_order_compute): Likewise.
24825 (pre_and_rev_post_order_compute_fn): Likewise.
24827 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24829 * genrecog.c (int_set::int_set): Explicitly construct our
24830 auto_vec base class.
24831 * vec.h (auto_vec::auto_vec): New constructor.
24833 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24835 * bitmap.h (class auto_bitmap): New constructor taking
24836 bitmap_obstack * argument.
24838 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24840 * bitmap.h (class auto_bitmap): Change type of m_bits to
24841 bitmap_head, and adjust ctor / dtor and member operators.
24843 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24845 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24846 when returned register mode doesn't match original mode.
24848 2017-05-12 Jeff Law <law@redhat.com>
24849 Jakub Jelinek <jakub@redhat.com>
24851 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24852 we look for cc setter after the compare-elim changes.
24853 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24854 within the vector to match what compare-elim now expects.
24855 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24856 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24858 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24859 after the compare-elim changes.
24860 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24861 the vector to match what compare-elim now expects.
24862 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24863 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24864 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24865 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24866 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24868 * config/visium/visium.c (single_set_and_flags): Fix where
24869 we look for cc setter after the compare-elim changes.
24870 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24871 with the vector to match what compare-elim now expects.
24872 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24873 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24874 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24875 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24876 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24877 (neg<mode>2_insn_set_overflow): Likewise.
24879 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24881 PR middle-end/79794
24882 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24883 maybe_expand_insn call, set ops[0].target. If still set after call,
24884 set alt_rtl. Add extra arg to recursive calls.
24885 (extract_bit_field): Add alt_rtl argument. Pass to
24887 * expmed.h (extract_bit_field): Fix prototype.
24888 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24889 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24890 to extract_bit_field_calls.
24891 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24892 Pass alt_rtl to extract_bit_field calls.
24893 * calls.c (store_unaligned_arguments_into_psuedos)
24894 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24895 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24897 * optabs.h (struct expand_operand): Add target bitfield.
24899 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24901 * compare-elim.c (try_eliminate_compare): Canonicalize
24902 operation with embedded compare to
24903 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24904 (set (reg) (operation)].
24906 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24908 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24911 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24912 cost of adding a carry flag for ADC instruction.
24913 [case MINUS]: Ignore the cost of subtracting a carry flag
24914 for SBB instruction.
24916 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24918 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24920 * config/rs6000/bmiintrin.h: New file.
24921 * config/rs6000/bmi2intrin.h: New file.
24922 * config/rs6000/x86intrin.h: New file.
24924 2017-05-12 Jeff Law <law@redhat.com>
24926 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24929 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24931 PR middle-end/80707
24932 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24934 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24936 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24937 early expansion of vector divide builtins.
24938 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24939 builtins identified as having unsigned arguments.
24941 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24943 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24944 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24945 expansion of vector logical operations (and, andc, or, xor,
24948 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24950 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24951 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24953 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24956 early GIMPLE expansion of vector multiplies.
24958 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24960 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24961 TARGET_HAVE_MOVT conditional.
24962 (movt splitter): Likewise.
24964 2017-05-12 Richard Biener <rguenther@suse.de>
24966 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24967 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24968 Fold all stmts not inplace.
24970 2017-05-12 Richard Biener <rguenther@suse.de>
24972 PR tree-optimization/80713
24973 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24974 inserted_exprs bit for not removed stmts.
24976 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24978 PR middle-end/69921
24979 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24980 parallelized" attribute for parallelized OpenACC kernels.
24981 * omp-offload.c (execute_oacc_device_lower): Use it.
24983 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24984 Set "oacc kernels" attribute.
24985 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24986 parameter. Adjust all users.
24987 (oacc_fn_attrib_kernels_p): Remove function.
24988 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24989 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24990 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24991 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24992 assert "oacc kernels" attribute is set.
24994 2017-05-11 Carl Love <cel@us.ibm.com>
24996 * config/rs6000/rs6000-c: Add support for built-in functions
24997 vector unsigned char vec_popcnt (vector signed char)
24998 vector unsigned char vec_popcnt (vector unsigned char)
24999 vector unsigned short vec_popcnt (vector signed short)
25000 vector unsigned short vec_popcnt (vector unsigned short)
25001 vector unsigned int vec_popcnt (vector signed int)
25002 vector unsigned int vec_popcnt (vector unsigned int)
25003 vector unsigned long long vec_popcnt (vector signed long long)
25004 vector unsigned long long vec_popcnt (vector unsigned long long)
25005 vector signed long long vec_slo (vector signed long long,
25006 vector signed char)
25007 vector signed long long vec_slo (vector signed long long,
25008 vector unsigned char)
25009 vector unsigned long long vec_slo (vector unsigned long long,
25010 vector signed char)
25011 vector unsigned long long vec_slo (vector unsigned long long,
25012 vector unsigned char)
25013 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25014 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25015 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25016 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25017 * doc/extend.texi: Update the built-in documentation file for the
25018 new built-in functions.
25020 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25022 * attribs.h (sorted_attr_string): Move machine independent
25023 functions for target clone support from the i386 port to common
25024 code. Rename ix86_function_versions to common_function_versions.
25025 Rename make_name to make_unique_name.
25026 (common_function_versions): Likewise.
25027 (make_unique_name): Likewise.
25028 (make_dispatcher_decl): Likewise.
25029 (is_function_default_version): Likewise.
25030 * attribs.c (attr_strcmp): Likewise.
25031 (sorted_attr_string): Likewise.
25032 (common_function_versions): Likewise.
25033 (make_unique_name): Likewise.
25034 (make_dispatcher_decl): Likewise.
25035 (is_function_default_version): Likewise.
25036 * config/i386/i386.c (attr_strcmp): Likewise.
25037 (sorted_attr_string): Likewise.
25038 (ix86_function_versions): Likewise.
25039 (make_name): Likewise.
25040 (make_dispatcher_decl): Likewise.
25041 (is_function_default_version): Likewise.
25042 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25044 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25047 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25048 Account for direct move costs for vec_construct of integer
25051 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25054 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25055 (UNSPEC_STX_ATOMIC): Ditto.
25056 (loaddi_via_sse): New insn.
25057 (storedi_via_sse): Ditto.
25058 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25059 Update corresponding peephole2 patterns.
25060 (atomic_storedi_fpu): Ditto.
25062 2017-05-11 Julia Koval <julia.koval@intel.com>
25064 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25065 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25067 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25068 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25069 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25071 2017-05-11 Nathan Sidwell <nathan@acm.org>
25073 * graphite-poly.c: Include dumpfile.h.
25075 * dumpfle.h (dump_function): Declare here ...
25076 * tree-dump.h (dump_function): ... not here.
25077 * dumpfile.c: #include tree-cfg.h.
25078 (dump_function): Move here from ...
25079 * tree-dump.c (dump_function): ... here.
25080 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25081 * graphite-poly.c: Don't include tree-dump.h.
25082 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25083 * print-tree.c: Likewise.
25084 * stor-layout.c: Likewise.
25085 * tree-nested.c: Likewise.
25087 * dumpfile.c (dump_start): Use TDF_FLAGS.
25088 (dump_enable_all): Fix TDF_KIND check thinko.
25090 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25092 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25093 array entries to represent two legal parameterizations of the
25094 overloaded __builtin_cmpb function, as represented by the
25095 P6_OV_BUILTIN_CMPB constant.
25096 (altivec_resolve_overloaded_builtin): Add special case handling
25097 for the __builtin_cmpb function, as represented by the
25098 P6_OV_BUILTIN_CMPB constant.
25099 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25100 (BU_P6_64BIT_2): New macro.
25101 (BU_P6_OVERLOAD_2): New macro
25102 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25103 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25104 (CMPB): Add overload support to represent both 32-bit and 64-bit
25105 compare-bytes function.
25106 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25107 support for TARGET_CMPB.
25108 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25109 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25110 documentation of the __builtin_cmpb overloaded built-in function.
25112 2017-05-11 Richard Biener <rguenther@suse.de>
25114 PR tree-optimization/80705
25115 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25116 bases are not vectorizable.
25118 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25120 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25121 when counting register pressure.
25123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25125 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25126 (iv_ca_more_deps): Renamed to ...
25127 (iv_ca_compare_deps): ... this.
25128 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25132 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25134 (determine_group_iv_costs): ... here.
25135 (find_inv_vars_cb): Record inv var if it's not recorded before.
25137 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25139 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25140 (get_shiftadd_cost): Ditto.
25142 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25144 * tree-ssa-address.c: Include header file.
25145 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25147 (add_to_parts): Refactor.
25148 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25149 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25152 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25154 PR tree-optimization/53090
25155 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25157 (extract_cond_operands): Detect condition with IV on both sides
25158 and return COMP_IV_EXPR_2.
25159 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25160 (rewrite_use_compare): Simplify by removing call to function
25161 extract_cond_operands.
25163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25165 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25166 (extract_cond_operands): Detect condition comparing against non-
25167 invariant bound and return appropriate enum value.
25168 (find_interesting_uses_cond): Update use of extract_cond_operands.
25169 Handle its return value accordingly.
25170 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25172 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25174 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25175 nonlinear iv_use computation in loop invariant sensitive way.
25177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25179 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25180 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25182 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25184 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25185 (dump_cand): Support iv_cand.inv_exprs.
25186 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25188 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25193 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25195 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25196 as local function. Include necessary header files.
25197 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25199 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25201 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25205 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25206 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25207 RSHIFT_EXPR and BIT_NOT_EXPR.
25209 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25211 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25212 (adjust_setup_cost): New parameter supporting round up adjustment.
25213 (struct address_cost_data): Delete.
25214 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25215 (split_address_cost, ptr_difference_cost): Delete.
25216 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25217 (struct ainc_cost_data): New struct.
25218 (get_address_cost_ainc): New function.
25219 (get_address_cost, get_computation_cost): Reimplement.
25220 (determine_group_iv_cost_address): Record inv_expr for all uses of
25222 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25223 (iv_ca_has_deps): Reimplemented to ...
25224 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25226 (iv_ca_extend): Call iv_ca_more_deps.
25228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25230 * tree-ssa-address.c (struct mem_address): Move to header file.
25231 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25232 * tree-ssa-address.h (struct mem_address): Move from C file.
25233 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25235 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25237 * tree-affine.h (aff_combination_type): New interface.
25238 (aff_combination_zero_p): Remove static.
25239 (aff_combination_const_p): New interface.
25240 (aff_combination_singleton_var_p): New interfaces.
25242 2017-05-11 Richard Biener <rguenther@suse.de>
25244 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25245 Skip unreachable blocks and destinations.
25246 (eliminate): Move stmt removal and fixup ...
25247 (fini_eliminate): ... here. Skip inserted exprs.
25248 (pass_pre::execute): Move fini_pre after fini_eliminate.
25249 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25250 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25251 PRE to get rid of dead code that has invalid SSA form and
25252 split critical edges again.
25254 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25256 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25258 2017-05-11 Richard Biener <rguenther@suse.de>
25260 * passes.c (execute_function_todo): Verify loops if they are
25261 said to be up-to-date.
25262 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25263 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25265 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25268 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25269 handle calling assemble_external ourself.
25272 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25273 modes with zero size. Enhance comment.
25275 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25277 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25278 built-ins for vec_xl and vec_xst with short and char pointer
25281 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25283 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25284 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25285 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25286 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25287 (_mm_maskz_min_round_ss): New intrinsics.
25288 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25289 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25290 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25291 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25292 (__builtin_ia32_minss_mask_round): New builtins.
25293 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25294 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25295 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25297 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25298 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25300 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25303 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25305 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25306 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25307 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25308 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25309 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25310 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25311 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25312 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25313 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25314 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25315 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25316 (__builtin_ia32_mulss_mask_round): New builtins.
25317 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25318 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25319 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25321 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25322 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25324 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25327 2017-05-10 Julia Koval <julia.koval@intel.com>
25329 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25330 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25331 (_mm256_setr_m128i): New intrinsics.
25333 2017-05-10 Julia Koval <julia.koval@intel.com>
25335 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25336 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25337 (_mm_maskz_rcp14_ss): New intrinsics.
25338 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25339 (__builtin_ia32_rcp14ss_mask): New builtins.
25340 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25342 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25344 PR tree-optimization/51513
25345 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25346 (assert_unreachable_fallthru_edge_p): Use it.
25347 (group_case_labels_stmt): Likewise.
25348 * tree-cfg.h: Prototype it.
25349 * stmt.c: Include cfghooks.h and tree-cfg.h.
25350 (emit_case_dispatch_table) <gap_label>: New local variable.
25351 Use it to fill dispatch table gaps.
25352 Test for default_label before updating probabilities.
25353 (expand_case) <default_label>: Remove unneeded initialization.
25354 Test for unreachable default case statement and remove its edge.
25355 Set default_label accordingly.
25356 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25358 2017-05-10 Carl Love <cel@us.ibm.com>
25360 * config/rs6000/rs6000-c: Add support for built-in functions
25361 vector signed char vec_neg (vector signed char)
25362 vector signed short int vec_neg (vector short int)
25363 vector signed int vec_neg (vector signed int)
25364 vector signed long long vec_neg (vector signed long long)
25365 vector float vec_neg (vector float)
25366 vector double vec_neg (vector double)
25367 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25369 * config/rs6000/altivec.h: Add define for vec_neg
25370 * doc/extend.texi: Update the built-in documentation for the
25371 new built-in functions.
25373 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25375 PR tree-optimization/77644
25376 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25378 2017-05-10 Nathan Sidwell <nathan@acm.org>
25380 * dumpfile.h (TDI_lang_all): New.
25381 (TDF_KIND): New. Renumber others
25382 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25384 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25386 (get_dump_file_name): Adjust suffix generation.
25387 (dump_enable_all): Use TDF_KIND.
25388 * doc/invoke.texi (-fdump-lang-all): Document.
25390 * dumpfile.h: Tabify.
25392 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25395 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25396 Move member access before delete.
25398 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25400 * tree-inline.c (expand_call_inline): Split block at stmt
25403 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25406 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25407 are now unused after splitting mov{sf,sd}_hardfloat.
25408 (f32_lr2): Likewise.
25409 (f32_lm): Likewise.
25410 (f32_lm2): Likewise.
25411 (f32_li): Likewise.
25412 (f32_li2): Likewise.
25413 (f32_lv): Likewise.
25414 (f32_sr): Likewise.
25415 (f32_sr2): Likewise.
25416 (f32_sm): Likewise.
25417 (f32_sm2): Likewise.
25418 (f32_si): Likewise.
25419 (f32_si2): Likewise.
25420 (f32_sv): Likewise.
25421 (f32_dm): Likewise.
25422 (f32_vsx): Likewise.
25423 (f32_av): Likewise.
25424 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25425 For movsf, order stores so the VSX stores occur before the GPR
25426 store which encourages the register allocator to use a traditional
25427 FPR instead of a GPR. For movsd, order the stores so that the GPR
25428 store comes before the VSX stores to allow the power6 to work.
25429 This is due to the power6 not having a 32-bit integer store
25430 instruction from a FPR.
25431 (movsf_hardfloat): Likewise.
25432 (movsd_hardfloat): Likewise.
25434 2017-05-09 Martin Sebor <msebor@redhat.com>
25436 PR translation/80280
25437 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25440 PR translation/80280
25441 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25442 data member added in r247778.
25443 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25445 2017-05-09 Nathan Sidwell <nathan@acm.org>
25447 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25449 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25452 2017-05-09 Marek Polacek <polacek@redhat.com>
25454 * doc/invoke.texi: Fix typo.
25456 2017-05-09 Richard Biener <rguenther@suse.de>
25458 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25459 (vrp_val_is_min): Likewise.
25460 (set_value_range_to_value): Likewise.
25461 (set_value_range_to_nonnegative): Likewise.
25462 (gimple_assign_nonzero_p): Likewise.
25463 (gimple_stmt_nonzero_p): Likewise.
25464 (vrp_int_const_binop): Likewise. Remove unreachable case.
25465 (adjust_range_with_scev): Adjust comments.
25466 (compare_range_with_value): Likewise.
25467 (extract_range_from_phi_node): Likewise.
25468 (test_for_singularity): Likewise.
25470 2017-05-09 Richard Biener <rguenther@suse.de>
25472 * tree-vrp.c (get_single_symbol): Add assert that we don't
25473 get overflowed constants as invariant part.
25474 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25475 checks. Use wi::cmp instead of recursing for integer constants.
25476 (compare_values): Just ignore whether we assumed undefined
25477 overflow instead of failing the compare.
25478 (extract_range_for_var_from_comparison_expr): Add comment before the
25479 TREE_NO_WARNING sets.
25480 (test_for_singularity): Likewise.
25481 (extract_range_from_comparison): Do not disable optimization
25482 when we assumed undefined overflow.
25483 (extract_range_basic): Remove init of unused var.
25485 2017-05-09 Richard Biener <rguenther@suse.de>
25487 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25488 (extract_range_from_multiplicative_op_1): Adjust.
25489 (extract_range_from_binary_expr_1): Use int_const_binop.
25491 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25494 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25495 rs6000_store_data_bypass_p in seven define_bypass directives and
25496 in several comments.
25497 * config/rs6000/rs6000-protos.h: Add prototype for
25498 rs6000_store_data_bypass_p function.
25499 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25500 function implements slightly different (rs6000-specific) semantics
25501 than store_data_bypass_p, returning false rather than aborting
25502 with assertion error when arguments do not satisfy the
25503 requirements of store data bypass.
25504 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25505 rs6000_store_data_bypass_p.
25507 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25509 * config/xtensa/xtensa-protos.h
25510 (xtensa_initial_elimination_offset): New declaration.
25511 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25512 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25513 macro definition, add case for FRAME_POINTER_REGNUM when
25514 FRAME_GROWS_DOWNWARD.
25515 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25516 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25517 xtensa_initial_elimination_offset.
25519 2017-05-08 Nathan Sidwell <nathan@acm.org>
25521 * doc/invoke.texi: Alphabetize -fdump options.
25523 2017-05-08 Martin Sebor <msebor@redhat.com>
25525 PR translation/80280
25526 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25528 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25530 * target.def (compute_frame_layout): New optional target hook.
25531 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25532 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25533 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25535 * reload1.c (verify_initial_elim_offsets): Likewise.
25536 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25537 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25538 (arm_get_frame_offsets): Split up into this ...
25539 (arm_compute_frame_layout): ... and this function.
25541 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25543 * config/aarch64/constraints.md (Usa): New constraint.
25544 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25546 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25548 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25549 with_multilib_list after it has been checked.
25551 2017-05-08 Richard Biener <rguenther@suse.de>
25553 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25554 (bitmap_set_subtract_values): Likewise.
25556 2017-05-08 Richard Biener <rguenther@suse.de>
25558 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25559 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25561 (gimple_stmt_nonzero_warnv_p): Rename to ...
25562 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25564 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25565 (extract_range_basic): Adjust, do not disable propagation on
25566 strict overflow sensitive simplification.
25567 (vrp_visit_cond_stmt): Likewise.
25569 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25571 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25572 body size unconditionally.
25574 2017-05-07 Jeff Law <law@redhat.com>
25577 2017-05-06 Jeff Law <law@redhat.com>
25578 PR tree-optimization/78496
25579 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25582 PR tree-optimization/78496
25583 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25584 (simplify_stmt_using_ranges): Call it.
25585 (vrp_dom_walker::before_dom_children): Extract equivalences
25586 from an ASSERT_EXPR with an equality comparison against a
25589 2017-05-06 Jeff Law <law@redhat.com>
25591 PR tree-optimization/78496
25592 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25595 PR tree-optimization/78496
25596 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25597 (simplify_stmt_using_ranges): Call it.
25598 (vrp_dom_walker::before_dom_children): Extract equivalences
25599 from an ASSERT_EXPR with an equality comparison against a
25602 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25604 * lra-constraints.c (lra_copy_reg_equiv): New function.
25605 (split_reg): Use it to copy equivalence information from the
25606 original register to the spill register.
25608 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25610 PR rtl-optimization/75964
25611 * simplify-rtx.c (simplify_const_relational_operation): Remove
25612 invalid handling of comparisons of integer ABS.
25614 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25616 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25617 initialize to zero.
25618 (init_regs): Remove declaration.
25619 (function_arg_advance_32): Initialize error_p as boolean variable.
25621 2017-05-05 Nathan Sidwell <nathan@acm.org>
25623 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25624 lines. Use for (;;).
25626 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25628 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25629 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25630 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25631 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25632 VF=2 that require versioning.
25634 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25636 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25639 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25641 * diagnostic.h (diagnostic_override_option_index): Convert from
25642 macro to inline function.
25644 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25646 * diagnostic.c (last_module_changed_p): New function.
25647 (set_last_module): New function.
25648 (diagnostic_report_current_module): Convert macro usage to
25649 the above functions.
25650 * diagnostic.h (diagnostic_context::last_module): Strengthen
25651 from const line_map * to const line_map_ordinary *.
25652 (diagnostic_last_module_changed): Delete macro.
25653 (diagnostic_set_last_module): Delete macro.
25655 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25657 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25658 with diagnostic_report_diagnostic.
25659 (diagnostic_n_impl_richloc): Likewise.
25660 * diagnostic.h (report_diagnostic): Delete macro.
25661 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25662 with diagnostic_report_diagnostic.
25663 * substring-locations.c (format_warning_va): Likewise.
25665 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25667 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25668 save/restor of format_spec. Move option-printing code to...
25669 (print_option_information): ...this new function, and
25670 reimplement by simply printing to the pretty_printer,
25671 rather than appending to the format string.
25673 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25675 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25676 handling logic into...
25677 (update_effective_level_from_pragmas): ...this new function.
25679 2017-05-04 Andrew Waterman <andrew@sifive.com>
25681 * config/riscv/riscv.opt (mstrict-align): New option.
25682 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25683 (SLOW_UNALIGNED_ACCESS): Define.
25684 (riscv_slow_unaligned_access): Declare.
25685 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25687 (riscv_slow_unaligned_access): New variable.
25688 (rocket_tune_info): Set slow_unaligned_access to true.
25689 (optimize_size_tune_info): Set slow_unaligned_access to false.
25690 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25691 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25692 (riscv_option_override): Set riscv_slow_unaligned_access.
25693 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25695 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25697 * config/riscv/riscv.md: Unify indentation.
25699 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25704 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25706 (extendsi<mode>2): Add support for doing sign extension via
25707 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25708 don't have ISA 3.0 instructions.
25709 (extendsi<mode>2 splitter): Likewise.
25710 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25711 generate the normal insns since SImode can now go in vector
25712 registers. Disallow the special UNSPECs needed for previous
25713 machines to hide SImode being used. Add new insns
25714 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25715 (fix_trunc<mode>si2_stfiwx): Likewise.
25716 (fix_trunc<mode>si2_internal): Likewise.
25717 (fixuns_trunc<mode>si2): Likewise.
25718 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25719 (fctiw<u>z_<mode>_smallint): Likewise.
25720 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25721 of floating point to 32-bit integer from doing a direct move to
25722 the GPR registers to do a store.
25723 (fctiwz_<mode>): Break long line.
25725 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25727 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25728 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25729 (addr_list, addr_offset_valid_p): New.
25730 (split_address_groups): Check offset validity with above function.
25731 (gt-tree-ssa-loop-ivopts.h): Include header file.
25733 2017-05-05 Nathan Sidwell <nathan@acm.org>
25735 * config.gcc (arm*-*-*): Add missing 'fi'.
25737 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25739 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25740 included in -fopt-info does not matter.
25741 * doc/optinfo.texi (-fopt-info): Fix description of default
25742 behavour. Explicitly say order of options included in -fopt-info
25745 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25747 * config.gcc: Allow combinations of aprofile and rmprofile values for
25748 --with-multilib-list.
25749 * config/arm/t-multilib: New file.
25750 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25751 variables. Remove setting of ISA and floating-point ABI in
25752 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25753 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25754 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25755 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25757 * config/arm/t-rmprofile: Likewise except for the matches changes.
25758 * doc/install.texi (--with-multilib-list): Document the combination of
25759 aprofile and rmprofile values and warn about pitfalls in doing that.
25761 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25763 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25764 (movdi_aarch64): Likewise.
25766 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25768 PR tree-optimization/80632
25769 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25771 (build_arrays): Initialize it for virtual phis.
25772 (fix_phi_nodes): Use it for virtual phis.
25774 PR tree-optimization/80558
25775 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25776 [x, y] op z into [x op, y op z] for op & or | if conditions
25779 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25780 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25783 * config/arm/arm.md (use_literal_pool): Remove.
25784 (64-bit immediate split): No longer takes cost into consideration
25785 if arm_disable_literal_pool is enabled.
25786 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25787 used when arm_disable_literal_pool is enabled.
25788 (arm_max_const_double_inline_cost): Remove use of
25789 arm_disable_literal_pool.
25790 (push_minipool_fix): Add assert.
25791 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25792 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25793 (no_literal_pool_sf_immediate): New.
25795 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25797 PR tree-optimization/80613
25798 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25799 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25801 2017-05-05 Richard Biener <rguenther@suse.de>
25803 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25805 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25807 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25808 of this flag from insn conditions due to removal from r247495.
25810 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25812 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25814 (arm_early_store_addr_dep_ptr): Likewise.
25815 * config/arm/aarch-common-protos.h
25816 (arm_early_load_addr_dep_ptr): Add prototype.
25817 (arm_early_store_addr_dep_ptr): Likewise.
25818 * config/arm/cortex-a53.md: Add new bypasses.
25820 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25822 * tree.c (next_type_uid): Change type to unsigned.
25823 (type_hash_canon): Decrement back next_type_uid if
25824 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25825 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25828 2017-05-04 Martin Sebor <msebor@redhat.com>
25830 * builtins.c: Fix a trivial typo in a comment.
25832 PR middle-end/79234
25833 * builtins.c (check_sizes): Adjust to handle reading past the end.
25834 Avoid printing excessive upper bound of ranges. Use %E to print
25835 tree nodes instead of converting them to %wu.
25836 (expand_builtin_memchr): New function.
25837 (compute_dest_size): Rename...
25838 (compute_objsize): ...to this.
25839 (expand_builtin_memcpy): Adjust.
25840 (expand_builtin_mempcpy): Adjust.
25841 (expand_builtin_strcat): Adjust.
25842 (expand_builtin_strcpy): Adjust.
25843 (check_strncat_sizes): Adjust.
25844 (expand_builtin_strncat): Adjust.
25845 (expand_builtin_strncpy): Adjust and simplify.
25846 (expand_builtin_memset): Adjust.
25847 (expand_builtin_bzero): Adjust.
25848 (expand_builtin_memcmp): Adjust.
25849 (expand_builtin): Handle memcmp.
25850 (maybe_emit_chk_warning): Check strncat just once.
25852 2017-05-04 Martin Sebor <msebor@redhat.com>
25854 PR preprocessor/79214
25855 PR middle-end/79222
25856 PR middle-end/79223
25857 * builtins.c (check_sizes): Add inlining context and issue
25858 warnings even when -Wno-system-headers is set.
25859 (check_strncat_sizes): Same.
25860 (expand_builtin_strncat): Same.
25861 (expand_builtin_memmove): New function.
25862 (expand_builtin_stpncpy): Same.
25863 (expand_builtin): Handle memmove and stpncpy.
25865 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25867 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25868 which is not used any more.
25870 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25872 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25874 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25876 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25877 (cortexa53_tunings): Likewise.
25878 (cortexa57_tunings): Likewise.
25879 (cortexa72_tunings): Likewise.
25880 (cortexa73_tunings): Likewise.
25882 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25884 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25885 Set loop alignment to 8.
25887 2017-05-04 Martin Sebor <msebor@redhat.com>
25889 PR translation/80280
25890 * builtins.c (expand_builtin_object_size): Add missing quoting to
25891 %D and like directives.
25892 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25893 (hsa_type_for_tree_type): Same.
25894 (verify_function_arguments): Same.
25895 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25896 * varasm.c (get_section): Same.
25899 2017-05-04 Martin Sebor <msebor@redhat.com>
25901 PR translation/80280
25902 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25904 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25906 * config/aarch64/aarch64.c (generic_addrcost_table):
25907 Change HI/TI mode setting.
25909 2017-05-04 Martin Jambor <mjambor@suse.cz>
25911 PR tree-optimization/80622
25912 * tree-sra.c (comes_initialized_p): New function.
25913 (build_accesses_from_assign): Only set write lazily when
25914 comes_initialized_p is false.
25915 (analyze_access_subtree): Use comes_initialized_p.
25916 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25917 instead of testing for PARM_DECL.
25919 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25921 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25922 constraint on operand 0 to allow more general addressing modes.
25923 Adjust output template.
25924 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25926 * config/aarch64/aarch64-protos.h
25927 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25928 * config/aarch64/constraints.md (Dp): New address constraint.
25929 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25932 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25934 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25935 update use of estimate_ipcp_clone_size_and_time.
25936 (estimate_local_effects): Update use of
25937 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25938 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25939 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25940 Return nonspecialized time.
25942 2017-05-04 Richard Biener <rguenther@suse.de>
25944 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25945 for the last VUSE which def dominates the PHI. Directly call
25947 (get_continuation_for_phi_1): Remove.
25949 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25951 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25952 to explain the use of truncating division. Cap the number of
25953 iterations to the maximum given by nb_iterations_upper_bound,
25956 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25958 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25959 * configure: Regenerate.
25960 * config.in: Regenerate.
25961 * config/i386/driver-mingw32.c: new file.
25962 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25963 * config.host: Link driver-mingw32.o on MinGW host.
25964 * doc/install.texi: Document new --enable-mingw-wildcard configure
25967 2017-05-04 Marek Polacek <polacek@redhat.com>
25969 PR tree-optimization/80612
25970 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25972 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25973 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25975 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25976 (movt splitter): Likewise.
25977 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25978 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25979 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25980 block for Thumb-1 with MOVT.
25981 (thumb2_legitimate_address_p): Move code block ...
25982 (can_avoid_literal_pool_for_label_p): ... into this new function.
25983 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25985 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25986 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25987 "M-profile targets with the MOVT instruction".
25989 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25991 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25992 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25993 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25995 2017-05-04 Martin Liska <mliska@suse.cz>
25997 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25998 variable cond_code.
26000 2017-05-04 Richard Biener <rguenther@suse.de>
26002 * tree.c (array_at_struct_end_p): Handle arrays at struct
26003 end with flexarrays more conservatively. Refactor and treat
26004 arrays of arrays or aggregates more strict. Fix
26005 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26006 * tree.c (array_at_struct_end_p): Adjust prototype.
26007 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26008 * gimple-fold.c (get_range_strlen): Likewise.
26009 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26011 2017-05-04 Richard Biener <rguenther@suse.de>
26013 PR tree-optimization/31130
26014 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26016 (supports_overflow_infinity): Likewise.
26017 (is_negative_overflow_infinity): Likewise.
26018 (is_positive_overflow_infinity): Likewise.
26019 (is_overflow_infinity): Likewise.
26020 (stmt_overflow_infinity): Likewise.
26021 (overflow_infinity_range_p): Likewise.
26022 (usable_range_p): Remove as always returning true.
26023 (make_overflow_infinity): Remove.
26024 (negative_overflow_infinity): Likewise.
26025 (positive_overflow_infinity): Likewise.
26026 (avoid_overflow_infinity): Likewise.
26027 (set_value_range): Adjust accordingly.
26028 (set_value_range_to_nonnegative): Likewise, remove now unused
26029 overflow_infinity arg.
26030 (vrp_operand_equal_p): Adjust.
26031 (update_value_range): Likewise.
26032 (range_int_cst_singleton_p): Likewise.
26033 (operand_less_p): Likewise.
26034 (compare_values_warnv): Likewise.
26035 (extract_range_for_var_from_comparison_expr): Likewise.
26036 (vrp_int_const_binop): Likewise.
26037 (zero_nonzero_bits_from_vr): Likewise.
26038 (extract_range_from_multiplicative_op_1): Likewise.
26039 (extract_range_from_binary_expr_1): Likewise.
26040 (extract_range_from_unary_expr): Likewise.
26041 (extract_range_from_comparison): Likewise.
26042 (extract_range_basic): Likewise.
26043 (adjust_range_with_scev): Likewise.
26044 (compare_ranges): Likewise.
26045 (compare_range_with_value): Likewise.
26046 (dump_value_range): Likewise.
26047 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26049 (simplify_cond_using_ranges): Adjust.
26051 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26053 * brig-builtins.def: Added a builtin for class_f64.
26054 * builtin-types.def: Added a builtin type needed by class_f64.
26056 2017-05-03 Jason Merrill <jason@redhat.com>
26058 * timevar.def: Add TV_CONSTEXPR.
26060 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26062 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26064 2017-05-03 Martin Jambor <mjambor@suse.cz>
26066 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26067 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26068 * ipa-cp.c (ipcp_propagate_stage): Do not call
26069 ipa_update_after_lto_read.
26070 * ipa-inline.c (ipa_inline): Likewise.
26072 2017-05-03 Martin Jambor <mjambor@suse.cz>
26074 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26075 tag. Added a default constructor and a destructor.
26076 (ipa_edge_args_sum_t): New class;
26077 (ipa_edge_args_sum): Declare.
26078 (ipa_edge_args_vector): Remove declaration.
26079 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26080 (ipa_free_edge_args_substructures): Remove declaration.
26081 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26082 (ipa_edge_args_info_available_for_edge_p): Likewise.
26083 * ipa-prop.c (ipa_edge_args_vector): Removed.
26084 (edge_removal_hook_holder): Likewise.
26085 (edge_duplication_hook_holder): Likewise.
26086 (ipa_edge_args_sum): New variable.
26087 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26088 ipa_edge_args_vector.
26089 (ipa_free_edge_args_substructures): Likewise.
26090 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26091 ipa_edge_args_vector.
26092 (ipa_edge_removal_hook): Turned into method
26093 ipa_edge_args_sum_t::remove.
26094 (ipa_edge_duplication_hook): Turned into method
26095 ipa_edge_args_sum_t::duplicate.
26096 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26097 registering edge hooks.
26098 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26099 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26100 ipa_edge_args_sum instead of ipa_edge_args_vector.
26101 * ipa-profile.c (ipa_profile): Likewise.
26103 2017-05-03 Martin Jambor <mjambor@suse.cz>
26105 * symbol-summary.h (function_summary): New method exists.
26106 (function_summary::symtab_removal): Deallocate through release.
26107 (call_summary): New class.
26108 (gt_ggc_mx): New overload.
26109 (gt_pch_nx): Likewise.
26110 (gt_pch_nx): Likewise.
26112 2017-05-03 Jeff Law <law@redhat.com>
26114 PR tree-optimization/78496
26115 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26116 from simplify_cond_using_ranges. Split off code to walk
26117 backwards through casts into ...
26118 (simplify_cond_using_ranges_2): New function.
26119 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26120 (execute_vrp): After identifying jump threads, call
26121 simplify_cond_using_ranges_2.
26123 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26126 * ipa-inline.h (inline_summary): Add ctor.
26127 (create_ggc): Do not use ggc_cleared_alloc.
26129 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26130 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26132 * gcc.c (handle_braces): Support escaping in switch matching
26134 * doc/invoke.texi (Spec Files): Document it.
26135 Remove superfluous @code markup in items.
26137 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26139 * diagnostic-show-locus.c (struct column_range): New struct.
26140 (get_affected_columns): New function.
26141 (get_printed_columns): New function.
26142 (struct correction): New struct.
26143 (correction::ensure_capacity): New function.
26144 (correction::ensure_terminated): New function.
26145 (struct line_corrections): New struct.
26146 (line_corrections::~line_corrections): New dtor.
26147 (line_corrections::add_hint): New function.
26148 (layout::print_trailing_fixits): Reimplement in terms of the new
26150 (selftest::test_overlapped_fixit_printing): New function.
26151 (selftest::diagnostic_show_locus_c_tests): Call it.
26153 2017-05-03 Nathan Sidwell <nathan@acm.org>
26155 Canonicalize canonical type hashing
26156 * tree.h (type_hash_canon_hash): Declare.
26157 * tree.c (type_hash_list, attribute_hash_list): Move into
26158 type_hash_canon_hash.
26159 (build_type_attribute_qual_variant): Break out hash code calc into
26160 type_hash_canon_hash.
26161 (type_hash_canon_hash): New. Generic type hash computation.
26162 (build_range_type_1, build_array_type_1, build_function_type,
26163 build_method_type_directly, build_offset_type, build_complex_type,
26164 make_vector_type): Call it.
26166 2017-05-03 Richard Biener <rguenther@suse.de>
26168 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26169 When all DRs have unknown misaligned do not always peel
26170 when there is a store but apply the same costing model as if
26171 there were only loads.
26173 2017-05-03 Richard Biener <rguenther@suse.de>
26176 PR tree-optimization/80492
26177 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26178 compare_base_decls returning dont-know properly.
26180 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26182 * config/arm/iterators.md (CCSI): New mode iterator.
26183 (arch): New mode attribute.
26184 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26185 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26186 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26187 code iterator for success result mode.
26188 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26189 the corresponding new insn generators.
26191 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26194 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26195 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26197 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26199 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26200 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26201 (DDR_A): Wrap DDR argument in brackets.
26202 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26203 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26204 (DDR_REVERSED_P): Likewise.
26206 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26208 PR tree-optimization/79472
26209 * tree-switch-conversion.c (struct switch_conv_info): Add
26210 contiguous_range and default_case_nonstandard fields.
26211 (collect_switch_conv_info): Compute contiguous_range and
26212 default_case_nonstandard fields, don't clear final_bb if
26213 contiguous_range and only the default case doesn't have the required
26215 (check_all_empty_except_final): Set default_case_nonstandard instead
26216 of failing if contiguous_range and the default case doesn't have empty
26218 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26219 and only the default case doesn't have the required constants. Skip
26221 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26222 if default_case_nonstandard.
26223 (build_constructors): Build constant 1 just once. Assert that default
26224 values aren't inserted in between cases if contiguous_range. Skip
26226 (build_arrays): Skip virtual phis.
26227 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26228 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26229 Handle virtual phis.
26230 (gen_inbound_check): Handle default_case_nonstandard case.
26231 (process_switch): Adjust check_final_bb caller. Call
26232 gather_default_values with the first non-default case instead of
26233 default case if default_case_nonstandard.
26235 2017-05-02 Nathan Sidwell <nathan@acm.org>
26237 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26238 check. Fix formatting.
26240 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26242 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26243 errors when comparing specialized and unspecialized times.
26245 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26247 * diagnostic-show-locus.c
26248 (layout::should_print_annotation_line_p): Make private.
26249 (layout::print_annotation_line): Make private.
26250 (layout::annotation_line_showed_range_p): Make private.
26251 (layout::show_ruler): Make private.
26252 (layout::print_source_line): Make private. Pass in line and
26253 line_width, rather than calling location_get_source_line. Drop
26255 (layout::print_leading_fixits): New method.
26256 (layout::print_any_fixits): Rename to...
26257 (layout::print_trailing_fixits): ...this, and make private.
26258 Don't print newline fixits.
26259 (diagnostic_show_locus): Move logic for printing one row into...
26260 (layout::print_line): ...this new function. Move the
26261 location_get_source_line call and error-handling from
26262 print_source_line to here. Call print_leading_fixits, and rename
26263 print_any_fixits to print_trailing_fixits.
26264 (selftest::test_fixit_insert_containing_newline): Update now that
26265 newlines are partially supported.
26266 (selftest::test_fixit_insert_containing_newline_2): New test.
26267 (selftest::test_fixit_replace_containing_newline): Update comments.
26268 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26269 * edit-context.c (class added_line): New class.
26270 (class edited_line): Describe newline handling in comment.
26271 (edited_line::actually_edited_p): New method.
26272 (edited_line::print_content): Delete redundant decl.
26273 (edited_line::m_predecessors): New field.
26274 (edited_file::print_content): Call edited_line::print_content.
26275 (edited_file::print_diff): Update to support newlines.
26276 (edited_file::print_diff_hunk): Likewise.
26277 (edited_file::print_run_of_changed_lines): New function.
26278 (edited_file::print_diff_line): Convert to...
26279 (print_diff_line): ...this.
26280 (edited_file::get_effective_line_count): New function.
26281 (edited_line::edited_line): Initialize new field m_predecessors.
26282 (edited_line::~edited_line): Clean up m_predecessors.
26283 (edited_line::apply_fixit): Handle newlines.
26284 (edited_line::get_effective_line_count): New function.
26285 (edited_line::print_content): New function.
26286 (edited_line::print_diff_lines): New function.
26287 (selftest::test_applying_fixits_insert_containing_newline): New
26289 (selftest::test_applying_fixits_replace_containing_newline): New
26291 (selftest::insert_line): New function.
26292 (selftest::test_applying_fixits_multiple_lines): Add example of
26294 (selftest::edit_context_c_tests): Call the new tests.
26296 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26298 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26299 parameter cand. Update dump information.
26300 (get_computation_cost): Update uses.
26302 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26304 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26305 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26306 (get_computation_at, rewrite_use_address): Update use of
26307 get_computation_aff.
26309 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26311 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26312 (get_computation): Delete.
26313 (get_computation_cost): Implement like get_computation_cost_at.
26314 Use get_computation_at.
26315 (get_computation_cost_at): Delete.
26316 (rewrite_use_nonlinear_expr): Use get_computation_at.
26317 (rewrite_use_compare, remove_unused_ivs): Ditto.
26319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26321 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26323 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26325 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26326 (ivopts_global_cost_for_size): Rename parameter and update uses.
26327 (iv_ca_recount_cost): Update uses.
26328 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26329 candidates seperately in n_invs and n_cands.
26330 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26332 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26334 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26335 (find_inv_vars_cb): New.
26336 (find_depends): Renamed to ...
26337 (find_inv_vars): ... this.
26338 (add_candidate_1, force_var_cost): Call find_inv_vars.
26339 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26341 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26343 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26344 inv_vars. Add inv_exprs.
26345 (struct iv_cand): Rename depends_on to inv_vars.
26346 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26347 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26348 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26349 (dump_cand): Dump inv_vars.
26350 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26351 (record_invariant, find_depends, add_candidate_1): Ditto.
26352 (set_group_iv_cost, force_var_cost): Ditto.
26353 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26354 (get_computation_cost_at, get_computation_cost): Ditto.
26355 (determine_group_iv_cost_generic): Ditto.
26356 (determine_group_iv_cost_address): Ditto.
26357 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26358 (determine_group_iv_costs): Ditto.
26359 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26360 (iv_ca_set_remove_invariants): Renamed to ...
26361 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26362 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26363 (iv_ca_set_add_invariants): Renamed to ...
26364 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26365 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26366 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26367 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26368 (create_new_ivs): Remove useless dump.
26370 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26372 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26374 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26375 (iv_ca_set_no_cp, create_new_iv): Ditto.
26377 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26379 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26381 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26383 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26384 function tree_check2.
26386 2017-05-02 Martin Liska <mliska@suse.cz>
26388 * doc/gcov.texi: Add missing preposition.
26389 * gcov.c (function_info::function_info): Properly fill up
26390 all member variables.
26392 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26394 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26396 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26398 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26400 2017-05-02 Martin Liska <mliska@suse.cz>
26403 * lto-streamer-in.c (lto_read_tree_1): Remove
26404 LTO_STREAMER_DEBUG.
26405 * lto-streamer.c (struct tree_hash_entry): Likewise.
26406 (struct tree_entry_hasher): Likewise.
26407 (tree_entry_hasher::hash): Likewise.
26408 (tree_entry_hasher::equal): Likewise.
26409 (lto_streamer_init): Likewise.
26410 (lto_orig_address_map): Likewise.
26411 (lto_orig_address_get): Likewise.
26412 (lto_orig_address_remove): Likewise.
26413 * lto-streamer.h: Likewise.
26414 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26415 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26417 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26419 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26420 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26421 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26422 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26423 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26424 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26425 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26426 (mm_maskz_sub_ss): New intrinsics.
26427 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26428 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26429 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26430 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26431 (__builtin_ia32_subss_mask_round): New builtins.
26432 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26433 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26434 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26436 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26437 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26439 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26442 2017-05-02 Martin Jambor <mjambor@suse.cz>
26444 PR tree-optimization/78687
26445 * tree-sra.c (access): New field parent.
26446 (process_subtree_disqualification): New function.
26447 (disqualify_candidate): Call it.
26448 (build_accesses_from_assign): Reset write flag if creating an
26450 (build_access_subtree): Fill in parent field and also prpagate
26451 down grp_write flag.
26452 (create_artificial_child_access): New parameter set_grp_write, set
26453 grp_write to its value.
26454 (propagate_subaccesses_across_link): Also propagate grp_write flag
26456 (propagate_all_subaccesses): Push the closest parent back to work
26457 queue if add_access_to_work_queue returned true.
26459 2017-05-02 Richard Biener <rguenther@suse.de>
26461 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26462 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26463 -fstrict-overflow documentation.
26464 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26465 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26466 flag_strict_overflow.
26467 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26468 * lto-opts.c (lto_write_options): Do not stream it.
26469 * lto-wrapper.c (merge_and_complain): Do not handle it.
26470 * opts.c (default_options_table): Do not set -fstrict-overflow.
26471 (finish_options): Likewise do not clear it when sanitizing.
26472 * simplify-rtx.c (simplify_const_relational_operation): Do not
26473 test flag_strict_overflow.
26475 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26477 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26478 using enabled attribute.
26479 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26480 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26481 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26482 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26483 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26484 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26485 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26486 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26487 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26488 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26490 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26492 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26494 2017-05-02 Richard Biener <rguenther@suse.de>
26496 PR tree-optimization/80591
26498 2017-04-10 Richard Biener <rguenther@suse.de>
26500 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26503 2017-05-02 Richard Biener <rguenther@suse.de>
26505 PR tree-optimization/80549
26506 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26507 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26508 headers if they do not have a preheader.
26510 2017-05-02 Martin Liska <mliska@suse.cz>
26513 * common.opt: Fix typo.
26514 * doc/invoke.texi: Likewise.
26516 2017-05-01 Jan Beulich <jbeulich@suse.com>
26518 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26519 swapping, add (x,x,m,x,n) alternative.
26521 2017-05-01 Nathan Sidwell <nathan@acm.org>
26523 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26524 unnecessary unadjusted_alignment check.
26526 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26529 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26531 * gimplify.c (gimplify_cilk_detach): New function.
26532 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26533 * tree-core.h: Document EXPR_CILK_SPAWN.
26534 * tree.h (EXPR_CILK_SPAWN): Define.
26536 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26538 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26539 to use new fixit_hint representation, using the "replace" logic.
26540 (get_line_span_for_fixit_hint): Likewise.
26541 (layout::print_any_fixits): Likewise.
26542 (selftest::test_one_liner_many_fixits): Rename to...
26543 (selftest::test_one_liner_many_fixits_1): ...this, and update
26544 comment and expected output to reflect that the multiple fix-it
26545 hints are now consolidated into one insertion.
26546 (selftest::test_one_liner_many_fixits_2): New test.
26547 (selftest::test_diagnostic_show_locus_one_liner): Update for
26549 (selftest::test_fixit_consolidation): Update for fix-it API
26551 * diagnostic.c (print_parseable_fixits): Likewise.
26552 * edit-context.c (edited_line::m_line_events): Convert from
26553 auto_vec <line_event *> to auto_vec <line_event>.
26554 (class line_event): Convert from abstract base class to a concrete
26555 class, taking over the role of replace_event.
26556 (class insert_event): Delete.
26557 (class replace_event): Rename to class line_event. Convert to
26559 (edit_context::add_fixits): Reimplement.
26560 (edit_context::apply_insert): Delete.
26561 (edit_context::apply_replace): Rename to...
26562 (edit_context::apply_fixit): ...this. Convert to half-open range.
26563 (edited_file::apply_insert): Delete.
26564 (edited_file::apply_replace): Rename to...
26565 (edited_file::apply_fixit): ...this.
26566 (edited_line::~edited_line): Drop deletion of events.
26567 (edited_line::apply_insert): Delete.
26568 (edited_line::apply_replace): Rename to...
26569 (edited_line::apply_fixit): ...this. Convert to half-open range.
26570 Update for change to type of m_line_events.
26571 * edit-context.h (edit_context::apply_insert): Delete.
26572 (edit_context::apply_replace): Rename to...
26573 (edit_context::apply_fixit): ...this.
26575 2017-05-01 Martin Sebor <msebor@redhat.com>
26577 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26580 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26583 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26584 __get_cpuid_max returns 0.
26585 (__get_cpuid_count): Ditto.
26587 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26589 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26590 replacement expression is another instance of one of its arguments.
26592 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26595 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26596 check for stack push/pop autoinc.
26597 * config/i386/i386.c (ix86_agi_dependent): Return false
26598 if the only reason why modified_in_p returned true is that
26599 addr is SP based and set_insn is a push or pop.
26601 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26603 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26606 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26609 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26610 (account_size_time): Use two predicates - exec_pred and
26612 (evaluate_conditions_for_known_args): Compute both clause and
26614 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26615 (inline_summary_t::duplicate): Update.
26616 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26618 (compute_inline_parameters): Likewise.
26619 (estimate_edge_size_and_time): Update caluclation of time.
26620 (estimate_node_size_and_time): Compute both time and nonspecialized
26622 (estimate_ipcp_clone_size_and_time): Update.
26623 (inline_merge_summary): Update.
26624 (do_estimate_edge_time): Update.
26625 (do_estimate_edge_size): Update.
26626 (do_estimate_edge_hints): Update.
26627 (inline_read_section, inline_write_summary): Stream both new predicates.
26628 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26630 (compute_inlined_call_time): Cleanup.
26631 (big_speedup_p): Update.
26632 (edge_badness): Update.
26633 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26634 (size_time_entry): Replace predicate by exec_predicate and
26635 nonconst_predicate.
26636 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26637 (estimate_edge_time): Return also nonspec_time.
26638 (reset_edge_growth_cache): Update.
26640 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26642 PR rtl-optimization/80491
26643 * ifcvt.c (noce_process_if_block): When looking for x setter
26644 with missing else_bb, don't check only the insn right before
26645 cond_earliest, but look for the last insn that x is modified in
26646 within the same bb.
26648 PR rtl-optimization/80491
26649 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26651 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26653 PR tree-optimization/80487
26654 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26656 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26658 PR tree-optimization/79697
26659 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26660 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26661 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26663 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26664 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26666 2017-04-28 Martin Sebor <msebor@redhat.com>
26668 PR tree-optimization/80523
26669 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26670 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26672 (maybe_warn, format_directive, parse_directive): Use new functions.
26673 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26675 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26677 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26679 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26681 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26682 target_header_dir): Set correctly.
26683 * configure: Regenerated.
26684 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26685 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26686 instead of SYSTEM_HEADER_DIR.
26688 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26690 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26691 (estimate_local_effects): Likewise.
26692 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26693 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26694 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26695 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26696 do_estimate_edge_time, estimate_edge_time): Likewise.
26697 * ipa-inline-analysis.c (estimate_node_size_and_time,
26698 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26699 (estimate_time_after_inlining): Remove.
26701 2017-04-28 Martin Liska <mliska@suse.cz>
26703 * doc/gcov.texi: Enhance documentation of gcov.
26705 2017-04-28 Martin Liska <mliska@suse.cz>
26707 * doc/gcov.texi: Sort options in alphabetic order.
26708 * doc/gcov-dump.texi: Likewise.
26709 * doc/gcov-tool.texi: Likewise.
26710 * gcov.c (print_usage): Likewise.
26711 * gcov-dump.c (print_usage): Likewise.
26712 * gcov-tool.c (print_merge_usage_message): Likewise.
26713 (print_rewrite_usage_message): Likewise.
26714 (print_overlap_usage_message): Likewise.
26716 2017-04-28 Martin Liska <mliska@suse.cz>
26718 PR gcov-profile/53915
26719 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26721 2017-04-28 Martin Liska <mliska@suse.cz>
26723 PR gcov-profile/79891
26724 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26725 is marked by compiler as living on a line.
26726 (get_cycles_count): Remove usage of the union.
26727 (output_intermediate_file): Likewise.
26728 (find_source): Fix GNU coding style.
26729 (accumulate_line_counts): Remove old non-all block mode.
26730 (output_lines): Remove usage of the union.
26731 * profile.c (output_location): Include all BBs, even if
26732 belonging to a same line (and file) as a previous BB.
26734 2017-04-28 Martin Liska <mliska@suse.cz>
26736 * gcov.c (process_args): Handle new argument 'w'.
26737 (read_graph_file): Assign ID to BBs.
26738 (output_branch_count): Display BB # if verbose flag is set.
26739 (output_lines): Likewise for arcs.
26740 (print_usage): Add '--verbose' option help.
26741 * doc/gcov.texi: Document --verbose (-w) option.
26743 2017-04-28 Martin Liska <mliska@suse.cz>
26745 * gcov.c (struct block_location_info): New struct.
26746 (process_file): Fill up the new structure.
26747 (read_graph_file): Replace usage of encoding by the newly added
26749 (add_line_counts): Likewise.
26750 (accumulate_line_counts): Remove usage of the union.
26751 (function_info::function_info): New function.
26752 (function_info::~function_info): Likewise.
26753 (process_file): Call delete instead of release_function.
26754 (release_function): Release the function.
26755 (release_structures): Call delete instead of release_function.
26756 (solve_flow_graph): Replace usage of num_blocks.
26757 (find_exception_blocks): Likewise.
26758 (output_lines): Fix GNU coding style.
26760 2017-04-28 Martin Liska <mliska@suse.cz>
26763 * coverage.c (coverage_remove_note_file): New function.
26764 * coverage.h: Declare the function.
26765 * toplev.c (finalize): Clean if an error has been seen.
26767 2017-04-28 Martin Liska <mliska@suse.cz>
26769 PR gcov-profile/80031
26770 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26771 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26772 * gcov.c (read_graph_file): Read just number of blocks.
26773 * profile.c (branch_prob): Do not stream 0 flags per a basic
26776 2017-04-28 Martin Liska <mliska@suse.cz>
26778 * gcov-dump.c (tag_*): Add new argument to declarations.
26779 (dump_gcov_file): Likewise.
26780 (tag_blocks): Add and use new argument depth.
26781 (tag_arcs): Likewise.
26782 (tag_lines): Likewise.
26783 (tag_counters): Likewise.
26784 (tag_summary): Likewise.
26785 (dump_working_sets): Use depth to do a proper indentation.
26787 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26790 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26791 * symtab.c (symtab_node::debug_symtab): Move definition here.
26793 2017-04-28 Richard Biener <rguenther@suse.de>
26795 * lto-streamer.h (LTO_major_version): Bump to 7.
26797 2017-04-28 Richard Biener <rguenther@suse.de>
26799 * tree-vrp.c (assert_info): New struct.
26800 (add_assert_info): New helper.
26801 (register_edge_assert_for_2): Refactor to add asserts to a vector
26803 (register_edge_assert_for_1): Likewise.
26804 (register_edge_assert_for): Likewise.
26805 (finish_register_edge_assert_for): New helper actually registering
26806 asserts where live on edge.
26807 (find_conditional_asserts): Adjust.
26808 (find_switch_asserts): Likewise.
26809 (evrp_dom_walker::try_find_new_range): Generalize.
26810 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26812 2017-04-27 Marek Polacek <polacek@redhat.com>
26815 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26816 arg10 and arg11 to itype.
26818 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26820 * doc/extend.texi (Object Size Checking): Improve grammar.
26822 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26825 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26826 that the logic for permitting reciprocal estimates matches that
26829 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26832 * tree.c (type_cache_hasher::equal): Only compare
26833 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26834 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26835 non-aggregate element types.
26836 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26837 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26839 2017-04-27 Richard Biener <rguenther@suse.de>
26841 PR middle-end/80533
26842 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26843 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26844 keeping a reference to a trailing array.
26846 2017-04-27 Richard Biener <rguenther@suse.de>
26848 PR middle-end/80539
26849 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26850 being in loop-closed SSA form conservatively.
26851 (chrec_fold_multiply_poly_poly): Likewise.
26853 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26855 PR middle-end/79665
26856 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26857 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26859 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26862 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26863 (aarch64_function_arg_alignment): Return unsigned int again, but still
26864 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26865 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26866 Don't emit -Wpsabi note.
26867 (aarch64_function_arg_boundary): Likewise.
26868 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26871 2017-04-26 Nathan Sidwell <nathan@acm.org>
26873 * tree.h (crc32_unsigned_n): Declare.
26874 (crc32_unsigned, crc32_unsigned): Make inline.
26875 * tree.c (crc32_unsigned_bits): Replace with ...
26876 (crc32_unsigned_n): ... this.
26877 (crc32_unsigned, crc32_byte): Remove.
26878 (crc32_string): Remove unnecessary braces.
26880 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26882 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26883 * ipa-inline-analysis.c (MAX_TIME): Remove.
26884 (account_size_time): Use sreal for time.
26885 (dump_inline_summary): Update.
26886 (estimate_function_body_sizes): Update.
26887 (estimate_edge_size_and_time): Update.
26888 (estimate_calls_size_and_time): Update.
26889 (estimate_node_size_and_time): Update.
26890 (inline_merge_summary): Update.
26891 (inline_update_overall_summary): Update.
26892 (estimate_time_after_inlining): Update.
26893 (inline_read_section): Update.
26894 (inline_write_summary): Update.
26895 * ipa-inline.c (compute_uninlined_call_time): Update.
26896 (compute_inlined_call_time): Update.
26897 (recursive_inlining): Update.
26898 (inline_small_functions): Update.
26899 (dump_overall_stats): Update.
26900 * ipa-inline.h: Include sreal.h.
26901 (size_time_entry): Turn time to sreal.
26902 (inline_summary): Turn self_time nad time to sreal.
26904 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26906 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26908 (sreal::stream_out, sreal::stream_in): New.
26909 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26911 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26913 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26916 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26919 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26921 (dimode_scalar_chain::compute_convert_gain): Ditto.
26922 (dimode_scalar_chain::make_vector_copies): Ditto.
26923 (dimode_scalar_chain::convert_reg): Ditto.
26924 (dimode_scalar_chain::convert_insn): Ditto.
26925 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26926 (VI248_AVX512BW_1): New mode iterator.
26927 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26928 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26931 2017-04-25 Martin Sebor <msebor@redhat.com>
26933 PR tree-optimization/80497
26934 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26935 constants are representable in HOST_WIDE_INT.
26936 (parse_directive): Ditto.
26938 2017-04-25 Martin Sebor <msebor@redhat.com>
26941 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26942 (new_zero_array): Adjust signature.
26943 (dom_info::dom_init): Used unsigned rather that size_t.
26944 (dom_info::dom_info): Same.
26946 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26947 Jakub Jelinek <jakub@redhat.com>
26950 * config/arm/arm.c: Include gimple.h.
26951 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26952 returns negative, increment ncrn only if it returned positive.
26953 (arm_needs_doubleword_align): Return int instead of bool,
26954 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26955 members, but if there is any such non-FIELD_DECL
26956 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26957 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26958 returns negative, increment nregs only if it returned positive.
26959 (arm_setup_incoming_varargs): Likewise.
26960 (arm_function_arg_boundary): Emit -Wpsabi note if
26961 arm_needs_doubleword_align returns negative, return
26962 DOUBLEWORD_ALIGNMENT only if it returned positive.
26964 2017-04-25 Marek Polacek <polacek@redhat.com>
26967 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26968 first argument to type.
26970 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26973 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26974 type checks to test for compatibility instead of equality.
26976 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26977 Jakub Jelinek <jakub@redhat.com>
26980 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26982 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26983 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26984 the alignment computation, but return their maximum in warn_alignment.
26985 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26986 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26988 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26989 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26992 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26994 * config/arc/simdext.md (dmpyh): Fix typo.
26996 2017-04-25 Richard Biener <rguenther@suse.de>
26998 PR tree-optimization/80492
26999 * alias.c (compare_base_decls): Handle registers with asm
27000 specification conservatively.
27001 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27002 compare_base_decls returning dont-know properly.
27004 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27006 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27007 (legitimate_offset_address_p): New function.
27008 (arc_legitimate_address_p): Use above function.
27010 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27012 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27014 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27016 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27017 ACCH registers whenever they are available.
27019 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27021 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27022 double regs fix when not used.
27024 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27026 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27028 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27029 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27031 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27033 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27034 when emitting short ADD instructions.
27036 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27038 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27040 (cmpsi_cc_c_insn): Likewise.
27041 (cbranchsi4_scratch): Compute proper instruction length using
27042 compact_hreg_operand.
27043 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27045 2017-04-25 Richard Biener <rguenther@suse.de>
27047 PR middle-end/80509
27048 * passes.c (pass_manager::pass_manager): Initialize
27049 m_name_to_pass_map.
27051 2017-04-25 Richard Biener <rguenther@suse.de>
27053 PR tree-optimization/79201
27054 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27056 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27059 * config/s390/vector.md: Split MEM->GPR vector moves for
27060 non-s_operand addresses.
27062 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27065 * config/s390/predicates.md (reload_const_wide_int_operand): New
27067 * config/s390/s390.md ("movti"): Remove d/P alternative.
27068 ("movti_bigconst"): New pattern definition.
27070 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27073 * s390-protos.h (s390_expand_cs_hqi): Removed.
27074 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27075 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27076 modes as well as CCZ1mode and CCZmode.
27077 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27078 signature of s390_emit_compare_and_swap.
27079 (s390_expand_cs_hqi): Likewise, make static.
27080 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27081 compare-and-swap, in some cases.
27082 (s390_expand_cs): Wrapper function.
27083 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27085 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27086 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27087 patterns for small and large integers. Forbid symref memory operands.
27088 Move expander to s390.c. Require cc register.
27089 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27090 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27091 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27092 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27093 symref memory operands. Remove CC mode and call s390_match_ccmode
27095 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27097 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27099 * config/s390/s390.md (define_peephole2): New peephole to help
27100 combining the load-and-test pattern with volatile memory.
27102 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27104 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27105 with CCZmode for TARGET_Z196.
27107 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27109 PR rtl-optimization/80501
27110 * combine.c (make_compound_operation_int): Set subreg_code to SET
27111 even for AND with mask of the sign bit of mode.
27113 PR rtl-optimization/80500
27114 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27115 sum's initial value.
27117 2017-04-25 Julian Brown <julian@codesourcery.com>
27118 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27120 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27122 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27124 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27126 2017-04-25 Julian Brown <julian@codesourcery.com>
27127 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27129 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27130 (thunderx2t99_sha): New Reservation.
27132 2017-04-25 Julian Brown <julian@codesourcery.com>
27133 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27135 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27136 type for 1-element load.
27138 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27140 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27142 2017-04-24 Martin Jambor <mjambor@suse.cz>
27144 PR tree-optimization/80293
27145 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27146 char arrays not totally scalarizable if it is false.
27147 (analyze_all_variable_accesses): Pass correct value in the new
27148 parameter. Add a statistics counter.
27150 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27152 PR middle-end/79931
27153 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27155 2017-04-24 Richard Biener <rguenther@suse.de>
27157 PR tree-optimization/80494
27158 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27159 out for complex types.
27161 2017-04-24 Richard Biener <rguenther@suse.de>
27163 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27164 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27165 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27166 (DFS): Adjust and never fail.
27167 (sccvn_dom_walker::fail): Remove.
27168 (sccvn_dom_walker::before_dom_children): Adjust.
27169 (run_scc_vn): Likewise and never fail.
27170 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27171 (pass_fre::execute): Likewise.
27173 2017-04-24 Richard Biener <rguenther@suse.de>
27175 PR tree-optimization/79725
27176 * tree-ssa-sink.c (statement_sink_location): Return whether
27177 failure reason was zero uses. Move that check later.
27178 (sink_code_in_bb): Deal with zero uses by removing the stmt
27181 2017-04-24 Richard Biener <rguenther@suse.de>
27184 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27185 pointer-based references.
27187 2017-04-24 Richard Biener <rguenther@suse.de>
27190 * pass_manager.h (pass_manager::operator new): Remove.
27191 (pass_manager::operator delete): Likewise.
27192 * passes.c (pass_manager::operator new): Remove.
27193 (pass_manager::operator delete): Likewise.
27194 (pass_manager::pass_manager): Zero individual pass members.
27196 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27199 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27200 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27201 Check "XEXP (src, 1)" operand here.
27202 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27203 Check "XEXP (src, 1)" operand here.
27204 (dimode_scalar_chain::make_vector_copies): Detect count register
27205 of a shift instruction. Zero extend count register from QImode
27206 to DImode to satisfy vector shift pattern count operand predicate.
27207 Substitute vector shift count operand with a DImode copy.
27208 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27211 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27213 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27214 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27215 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27216 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27217 (UNSPEC_NOREX_MEM): Remove definition.
27219 2017-04-21 Richard Biener <rguenther@suse.de>
27221 PR tree-optimization/79547
27222 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27223 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27224 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27225 without any constraints.
27227 2017-04-21 Richard Biener <rguenther@suse.de>
27229 PR tree-optimization/78847
27230 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27232 2017-04-21 Richard Biener <rguenther@suse.de>
27234 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27235 (build_distinct_type_copy): Likewise.
27236 (build_variant_type_copy): Likewise.
27237 * tree.c (build_qualified_type): Pass down mem-stat info.
27238 (build_distinct_type_copy): Likewise.
27239 (build_variant_type_copy): Likewise.
27241 2017-04-21 Richard Biener <rguenther@suse.de>
27243 PR tree-optimization/80237
27244 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27246 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27247 for a simplified result.
27249 2016-04-21 Richard Biener <rguenther@suse.de>
27251 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27252 sth as strict as a simple_iv but a chrec without symbols and an
27253 operand defined in the loop we are peeling (and not some subloop).
27254 (propagate_constants_for_unrolling): Propagate all constants.
27256 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27259 * config/i386/i386.c (print_reg): Remove assert for disalowed
27260 regno values, call output_operand_lossage instead.
27262 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27265 * config/i386/constraints.md (Yc): New register constraint.
27266 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27267 Use Yc constraint for alternative 2 of operand 0. Remove
27268 preferred_for_speed attribute.
27270 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27272 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27273 lastprivate clauses in SIMT case.
27275 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27277 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27279 2017-04-20 Richard Biener <rguenther@suse.de>
27281 PR tree-optimization/57796
27282 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27283 as N scalar stores.
27284 (vect_model_load_cost): Cost gathers as N scalar loads.
27286 2017-04-20 Richard Biener <rguenther@suse.de>
27288 * ggc-page.c (ggc_allocated_p): Rename to ...
27289 (safe_lookup_page_table_entry): ... this and return the lookup
27291 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27293 2017-04-20 Richard Biener <rguenther@suse.de>
27295 PR tree-optimization/80453
27296 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27297 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27298 from the conditions.
27299 (vn_phi_eq): Pass them down.
27300 (vn_phi_lookup): Record them.
27301 (vn_phi_insert): Likewise.
27303 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27305 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27306 uninitialized variable warning to avoid buffer overrun.
27308 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27311 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27312 is suppressed for '{ 0 }' in C.
27314 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27316 * BASE-VER: Set to 8.0.0.
27318 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27320 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27321 priority .init_array and .fini_array section with SECTION_NOTYPE
27324 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27326 PR middle-end/80423
27327 * tree.h (build_array_type): Add typeless_storage default argument.
27328 * tree.c (type_cache_hasher::equal): Also compare
27329 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27330 (build_array_type): Add typeless_storage argument, set
27331 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27333 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27334 (build_array_type): Likewise. Add typeless_storage argument.
27336 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27337 Jakub Jelinek <jakub@redhat.com>
27339 PR tree-optimization/80426
27340 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27341 operation on symbolic operands, also compute the overflow for the
27342 invariant part when the operation degenerates into a negation.
27344 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27347 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27348 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27351 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27353 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27356 * config/avr/avr.c (tree.h): Include it.
27357 (cgraph.h): Include it.
27358 (avr_encode_section_info): Don't warn for uninitialized progmem
27359 variable if it's just an alias.
27361 2017-04-19 Richard Biener <rguenther@suse.de>
27364 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27365 when needed by AutoPGO.
27367 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27370 * doc/lto.texi: Remove an extra 'that'.
27372 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27374 PR rtl-optimization/80429
27375 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27376 are only used in debug insns.
27378 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27379 Vladimir Makarov <vmakarov@redhat.com>
27381 * config/sparc/predicates.md (input_operand): Add comment. Return
27382 true for any memory operand when LRA is in progress.
27383 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27385 2017-04-18 Jeff Law <law@redhat.com>
27388 * mips.md ({return,simple_return}_internal): Do not overwrite
27391 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27393 PR tree-optimization/80443
27394 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27395 instead of adding 1, subtract -1 and similarly instead of subtracting
27398 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27400 PR rtl-optimization/80357
27401 * haifa-sched.c (tmp_bitmap): New variable.
27402 (model_recompute): Handle duplicate use records.
27403 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27404 (free_global_sched_pressure_data): Free it.
27406 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27409 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27410 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27411 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27412 instead of SYSTEM_HEADER_DIR.
27414 2017-04-18 Jeff Law <law@redhat.com>
27416 PR middle-end/80422
27417 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27418 predecessors after walking up the insn chain.
27420 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27423 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27424 sizetype type into debug info.
27426 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27429 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27430 unneeded test for TARGET_UPPER_REGS_SF.
27431 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27433 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27436 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27437 instead of gsi_after_labels.
27439 2017-04-18 Jeff Law <law@redhat.com>
27441 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27445 2017-04-13 Jeff Law <law@redhat.com>
27446 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27447 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27449 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27452 * config/avr/avr.c (intl.h): Include it.
27453 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27455 2017-04-18 Martin Liska <mliska@suse.cz>
27457 PR gcov-profile/78783
27458 * gcov-tool.c (gcov_output_files): Validate that destination
27459 file is either removed by the tool or by a user.
27461 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27462 Guy Benyei <guybe@mellanox.com>
27464 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27465 block, and do not negate it, the stored id is already negative.
27467 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27469 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27471 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27474 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27475 masks of options that should be turned off if the VSX vector
27476 options are turned off.
27477 (OTHER_P8_VECTOR_MASKS): Likewise.
27478 (OTHER_VSX_VECTOR_MASKS): Likewise.
27479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27480 rs6000_disable_incompatible_switches to validate no type switches
27482 (rs6000_incompatible_switch): New function to disallow turning on
27483 other vector options if -mno-vsx, -mno-power8-vector, or
27484 -mno-power9-vector are specified.
27486 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27488 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27490 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27492 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27493 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27494 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27495 (ARG_POINTER_CFA_OFFSET): Likewise.
27497 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27499 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27500 conditions to take advantage of various optimizations.
27502 2017-04-13 Jeff Law <law@redhat.com>
27504 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27505 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27506 (zero_extendsidi2_dext): Likewise.
27508 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27511 * fold-const.c (fold_ternary_loc): Revert
27512 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27515 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27517 PR rtl-optimization/80343
27518 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27519 new scratch pseudo.
27521 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27524 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27525 to ubsan_encode_value.
27527 2017-04-13 Jeff Law <law@redhat.com>
27529 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27530 appearing in DEBUG_INSNs.
27532 2017-04-13 Martin Liska <mliska@suse.cz>
27534 PR gcov-profile/80413
27535 * gcov-io.c (gcov_write_string): Copy to buffer just when
27536 allocated size is greater than zero.
27538 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27541 * dwarf2out.c (decls_for_scope): Ignore declarations of
27542 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27544 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27547 * ipa-visibility.c (non_local_p): Fix typos.
27548 (localize_node): When localizing symbol in same comdat group,
27549 dissolve the group only when we know external symbols are going
27551 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27553 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27555 PR tree-optimization/79390
27556 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27557 order does not result in usable sequence, retry with reversed operand
27563 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27564 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27565 op0 instead of fold_convert_loc (loc, type, arg0).
27567 2017-04-12 Jeff Law <law@redhat.com>
27569 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27570 has a delay slot in the generated code.
27572 * config/cris/cris.md (cris_preferred_reload_class): Return
27573 GENNONACR_REGS rather than GENERAL_REGS.
27575 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27578 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27579 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27580 signedness of the result type.
27582 2017-04-12 Richard Biener <rguenther@suse.de>
27583 Jeff Law <law@redhat.com>
27585 PR tree-optimization/80359
27586 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27587 trim stores to TARGET_MEM_REFs.
27589 2017-04-12 Richard Biener <rguenther@suse.de>
27591 PR tree-optimization/79390
27592 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27593 threading case even more.
27595 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27598 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27599 for quad_address_p for TImode, instead of just not indexed_address.
27601 2017-04-12 Richard Biener <rguenther@suse.de>
27602 Bernd Edlinger <bernd.edlinger@hotmail.de>
27604 PR middle-end/79671
27605 * alias.c (component_uses_parent_alias_set_from): Handle
27606 TYPE_TYPELESS_STORAGE.
27607 (get_alias_set): Likewise.
27608 * tree-core.h (tree_type_common): Add typeless_storage flag.
27609 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27610 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27611 for types containing members with TYPE_TYPELESS_STORAGE.
27612 (place_field): Likewise.
27613 (layout_type): Likewise for ARRAY_TYPE.
27614 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27615 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27616 TYPE_TYPELESS_STORAGE.
27617 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27619 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27622 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27623 first argument to type.
27625 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27629 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27630 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27631 (rs6000_expand_binop_builtin): Likewise.
27632 (rs6000_expand_ternop_builtin): Likewise; also add missing
27633 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27634 vshasigma built-ins.
27635 * doc/extend.texi: Document that vec_xxpermdi's third argument
27636 must be a constant.
27638 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27640 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27641 Use shift_const cost parameter when calculating gain of STV shifts.
27643 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27645 PR rtl-optimization/70478
27646 * lra-constraints.c (process_alt_operands): Check memory for
27647 disfavoring memory insn operand.
27649 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27651 PR middle-end/80100
27652 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27653 left shift in unsigned HOST_WIDE_INT type.
27655 PR rtl-optimization/80385
27656 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27657 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27660 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27661 if they have any depend clauses.
27663 2017-04-11 Martin Liska <mliska@suse.cz>
27666 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27667 * ipa-split.c (split_function): Create a local comdat symbol
27668 if caller is in a comdat group.
27670 2017-04-11 Martin Liska <mliska@suse.cz>
27673 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27676 2017-04-11 Martin Sebor <msebor@redhat.com>
27678 PR middle-end/80364
27679 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27680 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27682 (directive::set_width, directive::set_precision, format_character):
27684 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27687 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27690 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27691 conflict, set target->arch_name instead of target->cpu_name.
27693 2017-04-11 Richard Biener <rguenther@suse.de>
27695 PR tree-optimization/80374
27696 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27697 build_zero_cst, remove fold_convertible_p check again.
27699 2017-04-11 Martin Liska <mliska@suse.cz>
27702 * ubsan.c (instrument_object_size): Do not instrument register
27705 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27708 * config/i386/i386-builtin-types.def
27709 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27710 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27711 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27712 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27713 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27714 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27715 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27716 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27717 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27718 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27719 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27720 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27721 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27722 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27723 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27724 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27725 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27726 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27727 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27728 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27729 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27730 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27731 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27732 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27733 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27734 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27735 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27736 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27737 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27738 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27739 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27740 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27741 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27742 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27743 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27744 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27745 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27746 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27747 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27748 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27749 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27750 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27751 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27752 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27753 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27754 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27756 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27757 flag to second_arg_count, handle 4 argument function type _COUNT
27758 aliases, handle second_arg_count on second argument rather than last.
27760 2017-04-10 Jeff Law <law@redhat.com>
27762 PR tree-optimization/80374
27763 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27764 record anything if we can not convert integer_zero_node to the
27767 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27770 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27771 Enhance special handling given to the TARGET_P9_MINMAX option in
27772 relation to certain other options.
27774 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27776 PR tree-optimization/80153
27777 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27778 remove POINTER_PLUS_EXPR's base part directly, rather than through
27781 2017-04-10 Richard Biener <rguenther@suse.de>
27782 Bin Cheng <bin.cheng@arm.com>
27784 PR tree-optimization/80153
27785 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27786 the first element of pointer type aff_tree. Build result expr in
27788 (add_elt_to_tree): Convert to type unconditionally. Remove other
27789 fold_convert calls.
27790 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27791 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27793 2017-04-10 Richard Biener <rguenther@suse.de>
27795 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27798 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27800 PR rtl-optimization/70478
27801 * lra-constraints.c (curr_small_class_check): New.
27802 (update_and_check_small_class_inputs): New.
27803 (process_alt_operands): Update curr_small_class_check. Disfavor
27804 alternative insn memory operands. Check available regs for small
27807 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27810 * config/mips/mips.opt (-mvirt): Update description.
27811 * doc/invoke.texi (-mvirt): Likewise.
27813 2017-04-10 Richard Biener <rguenther@suse.de>
27815 PR middle-end/80362
27816 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27817 looking for NEGATE_EXPR in -A / -B to A / B folding.
27819 2017-04-10 Martin Liska <mliska@suse.cz>
27821 PR gcov-profile/80224
27822 * gcov.c (print_usage): Fix usage string.
27823 (get_gcov_intermediate_filename): Remove.
27824 (output_gcov_file): Use both for normal and intermediate format.
27825 (generate_results): Do not initialize special file for
27826 intermediate format.
27828 2017-04-10 Richard Biener <rguenther@suse.de>
27830 PR tree-optimization/80304
27831 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27834 2017-04-10 Nathan Sidwell <nathan@acm.org>
27837 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27838 (rs6000_init_builtins): Use it.
27840 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27842 * config/arm/arm.md (<mrc>): Add mode to SET source.
27843 (<mrrc>): Likewise.
27845 2017-04-10 Richard Biener <rguenther@suse.de>
27847 PR middle-end/80344
27848 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27850 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27853 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27854 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27855 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27856 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27857 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27858 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27859 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27860 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27861 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27862 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27863 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27864 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27865 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27866 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27867 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27868 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27869 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27870 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27871 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27872 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27873 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27874 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27875 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27877 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27879 PR rtl-optimization/70478
27880 * lra-constraints.c: Reverse the last patch.
27882 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27884 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27885 Add comment for WCHAR_T.
27887 2017-04-08 Martin Liska <mliska@suse.cz>
27890 2017-04-07 Martin Liska <mliska@suse.cz>
27893 * ipa-split.c (split_function): Add function part to a same comdat
27896 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27899 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27901 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27903 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27904 (vec_pairing): ...to this.
27905 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27906 (rs6000_sched_init): Adjust for name change.
27907 (struct rs6000_sched_context): Likewise.
27908 (rs6000_init_sched_context): Likewise.
27909 (rs6000_set_sched_context): Likewise.
27911 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27917 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27919 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27920 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27921 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27923 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27925 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27927 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27929 PR rtl-optimization/70703
27930 * ira-color.c (update_conflict_hard_regno_costs): Use
27931 int64_t instead of HOST_WIDE_INT.
27933 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27935 PR rtl-optimization/70478
27936 * lra-constraints.c (process_alt_operands): Disfavor alternative
27937 insn memory operands.
27939 2017-04-07 Jeff Law <law@redhat.com>
27941 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27942 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27944 2017-04-07 Martin Liska <mliska@suse.cz>
27947 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27948 Show error message instead of an ICE.
27950 2017-04-07 Martin Liska <mliska@suse.cz>
27953 * ipa-split.c (split_function): Add function part to a same comdat
27956 2017-04-07 Richard Biener <rguenther@suse.de>
27958 PR middle-end/80341
27959 * tree.c (get_unwidened): Also handle ! for_type case for
27961 * convert.c (do_narrow): Split out from ...
27962 (convert_to_integer_1): ... here. Do not pass final truncation
27963 type to get_unwidened for TRUNC_DIV_EXPR.
27965 2017-04-07 Richard Biener <rguenther@suse.de>
27967 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27969 (aff_combination_const): Adjust.
27970 (aff_combination_scale): Likewise.
27971 (aff_combination_add_elt): Likewise.
27972 (aff_combination_add_cst): Likewise.
27973 (aff_combination_convert): Likewise.
27974 (add_elt_to_tree): Likewise. Remove unused argument.
27975 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27977 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27979 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27981 * config/arm/arm.c (arm_default_short_enums): Use
27982 ARM_DEFAULT_SHORT_ENUMS.
27983 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27985 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27988 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27989 members with redundant out-of-class redeclaration.
27991 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27994 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27995 * config/i386/i386.md (*zero_extendsidi2):
27996 Add (?*x,*x) and (?*v,*v) alternatives.
27998 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28001 * config/i386/i386.c (ix86_expand_builtin)
28002 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28003 mode from insn data. Convert operands to insn operand mode.
28004 Copy operands that don't satisfy insn predicate to a register.
28006 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28008 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28011 2017-04-06 Richard Biener <rguenther@suse.de>
28013 PR tree-optimization/80334
28014 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28015 preserve alignment of accesses.
28017 2017-04-06 Richard Biener <rguenther@suse.de>
28019 PR tree-optimization/80262
28020 * tree-sra.c (build_ref_for_offset): Preserve address-space
28022 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28023 Drop useless address-space information on MEM_REF offsets.
28025 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28027 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28029 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28031 PR rtl-optimization/70703
28032 * ira-color.c (update_conflict_hard_regno_costs): Use
28033 HOST_WIDE_INT instead of long.
28035 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28038 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28039 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28041 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28042 for x86_64 target. Handle -m3dnowa option.
28044 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28046 PR rtl-optimization/70703
28047 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28048 (update_conflict_hard_regno_costs): Use long instead of unsigned
28049 arithmetic for cost calculation.
28051 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28052 Bernd Edlinger <bernd.edlinger@hotmail.de>
28055 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28058 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28061 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28062 ptr_mode with Pmode throughout.
28063 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28064 into probe_stack_range and use DImode.
28066 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28069 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28070 call_eh_return is true.
28072 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28074 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28075 Initialize last_match_fntype_index.
28077 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28080 * tree-nvr.c: Include internal-fn.h.
28081 (pass_return_slot::execute): Ignore internal calls without
28084 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28085 Richard Biener <rguenther@suse.de>
28088 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28089 captures used multiple times, except for the last use.
28090 * generic-match-head.c: Include gimplify.h.
28092 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28094 PR tree-optimization/79390
28095 * target.h (struct noce_if_info): Declare.
28096 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28097 * target.def (noce_conversion_profitable_p): New target hook.
28098 * ifcvt.h (struct noce_if_info): New type, moved from ...
28099 * ifcvt.c (struct noce_if_info): ... here.
28100 (noce_conversion_profitable_p): Renamed to ...
28101 (default_noce_conversion_profitable_p): ... this. No longer
28103 (noce_try_store_flag_constants, noce_try_addcc,
28104 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28105 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28106 instead of noce_conversion_profitable_p.
28107 * config/i386/i386.c: Include ifcvt.h.
28108 (ix86_option_override_internal): Don't override
28109 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28110 (ix86_noce_conversion_profitable_p): New function.
28111 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28112 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28113 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28114 * doc/tm.texi: Regenerated.
28116 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28118 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28121 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28124 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28125 instructions for small multiply cores.
28127 2017-04-04 Jeff Law <law@redhat.com>
28129 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28131 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28132 that are not set by the loop over the elements.
28134 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28137 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28138 int mode, convert_modes it to mode as unsigned, otherwise use
28139 lowpart_subreg to mode rather than SImode.
28140 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28141 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28142 Use DImode instead of SImode for the shift count operand.
28143 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28146 2017-04-04 Richard Biener <rguenther@suse.de>
28148 PR middle-end/80281
28149 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28150 arithmetic done for the negate or the plus. Simplify.
28151 (A - (-B) -> A + B): Likewise.
28152 * fold-const.c (split_tree): Make sure to not negate pointers.
28154 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28156 PR rtl-optimization/60818
28157 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28158 a compare of comparisons with the thing compared if this results
28159 in a different machine mode.
28161 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28163 * alias.c (base_alias_check): Fix typo in comment.
28164 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28165 * cgraphunit.c (symbol_table::compile): Likewise.
28166 * collect2.c (maybe_run_lto_and_relink): Likewise.
28167 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28168 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28169 * config/avr/avr.c (avr_map_op_t): Likewise.
28170 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28171 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28172 * config/epiphany/epiphany.md (movcc): Likewise.
28173 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28174 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28176 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28177 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28178 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28179 * config/sh/sh.c (sh_rtx_costs): Likewise.
28180 * fold-const.c (fold_truth_andor): Likewise.
28181 * genautomata.c (collapse_flag): Likewise.
28182 * gengtype.h (struct type::u::s): Likewise.
28183 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28184 * input.c (FORMAT_AMOUNT): Likewise.
28185 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28186 (known_aggs_to_agg_replacement_list): Likewise.
28187 * ipa-inline-analysis.c: Likewise.
28188 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28189 * ipa-polymorphic-call.c
28190 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28191 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28192 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28194 * modulo-sched.c (apply_reg_moves): Likewise.
28195 * omp-expand.c (build_omp_regions_1): Likewise.
28196 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28197 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28198 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28199 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28200 * value-prof.c: Likewise.
28201 * var-tracking.c (val_reset): Likewise.
28203 2017-04-03 Richard Biener <rguenther@suse.de>
28205 PR tree-optimization/80275
28206 * fold-const.c (split_address_to_core_and_offset): Handle
28209 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28211 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28212 descriptors is at least equal to that of functions.
28214 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28216 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28218 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28221 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28222 (mov<IMOD4:mode>): New expander.
28223 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28225 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28227 PR rtl-optimization/79405
28228 * fwprop.c (propagations_left): New variable.
28229 (forward_propagate_into): Decrement it.
28230 (fwprop_init): Initialize it.
28231 (fw_prop): If the variable has reached zero, stop propagating.
28232 (fwprop_addr): Ditto.
28234 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28237 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28238 a FUNCTION_DECL, pass it as decl instead of origin to
28241 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28243 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28246 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28249 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28250 TARGET_VSX_SMALL_INTEGER.
28252 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28254 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28255 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28257 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28259 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28260 extraction from odd-numbered MSA register.
28262 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28264 PR middle-end/80173
28265 * expmed.c (store_bit_field_1): Don't attempt to create
28266 a word subreg out of hard registers wider than word if they
28267 have HARD_REGNO_NREGS of 1 for their mode.
28269 PR middle-end/80163
28270 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28271 conversions to integer types wider than word and pointer.
28274 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28275 (rtx_equal_for_cselib_p): Pass 0 to it.
28276 * cselib.c (cselib_hasher::equal): Likewise.
28277 (rtx_equal_for_cselib_1): Add depth argument. If depth
28278 is 128, don't look up VALUE locs and punt. Increment
28279 depth in recursive calls when walking VALUE locs.
28281 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28283 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28284 (make_gcov_file_name): Use the canonical path name for generating
28286 (read_line): Fix handling of files with ascii null bytes.
28288 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28290 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28291 to initialise a vector register instead
28292 of using a const_int.
28294 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28296 PR translation/80189
28297 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28298 diagnostic messages.
28300 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28303 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28304 (dfp_diex_<mode>): Update mode of operand 1.
28305 * doc/extend.texi (dxex, dxexq): Document change to return type.
28306 (diex, diexq): Document change to argument type.
28308 2017-03-30 Martin Jambor <mjambor@suse.cz>
28311 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28312 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28313 it reflects the signature changes performed at the callee side.
28314 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28315 to cgraph_build_function_type_skip_args.
28316 (build_function_decl_skip_args): Adjust call to the above function.
28318 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28321 * config/i386/sse.md
28322 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28323 register as dest whenever it is a MEM not rtx_equal_p to the
28324 corresponding dup operand, and when forcing into reg move the
28325 reg into the memory afterwards.
28326 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28327 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28328 for the force_reg mode.
28329 (avx512vl_vextractf128<mode>): Use register as dest either
28330 always when a MEM, or when it is a MEM not rtx_equal_p to the
28331 corresponding dup operand, or even not when it is a CONST_VECTOR
28332 depending on the mode and lo vs. hi.
28333 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28335 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28336 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28337 Likewise. Require that operands[2] is even.
28338 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28339 Remove extraneous parens. Require that operands[2] is a multiple
28341 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28342 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28343 disallow memory then.
28345 2017-03-30 Richard Biener <rguenther@suse.de>
28347 PR tree-optimization/77498
28348 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28349 to non-constants over backedges.
28351 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28353 PR rtl-optimization/80233
28354 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28355 as last_combined_insn. Do not test for BARRIER_P separately.
28357 2017-03-29 Andreas Schwab <schwab@suse.de>
28360 * calls.c (prepare_call_address): Convert funexp to Pmode before
28361 copying to temp reg.
28363 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28365 PR tree-optimization/80158
28366 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28367 Handle possible future case of more than one alternate
28369 (replace_rhs_if_not_dup): Likewise.
28370 (replace_one_candidate): Likewise.
28372 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28374 PR rtl-optimization/80193
28375 * ira.c (ira): Do not check allocation for LRA.
28377 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28379 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28380 (nvptx_output_simt_exit): Declare.
28381 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28382 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28383 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28384 (nvptx_declare_function_name): ...here. Emit declaration of local
28385 memory space buffer for omp_simt_enter insn.
28386 (nvptx_output_unisimt_switch): New.
28387 (nvptx_output_softstack_switch): New.
28388 (nvptx_output_simt_enter): New.
28389 (nvptx_output_simt_exit): New.
28390 * config/nvptx/nvptx.h (struct machine_function): New fields
28391 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28392 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28393 (UNSPECV_SIMT_EXIT): Ditto.
28394 (omp_simt_enter_insn): New insn.
28395 (omp_simt_enter): New expansion.
28396 (omp_simt_exit): New insn.
28397 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28399 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28400 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28401 (expand_GOMP_SIMT_EXIT): New.
28402 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28403 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28404 (GOMP_SIMT_EXIT): Ditto.
28405 * target-insns.def (omp_simt_enter): New insn.
28406 (omp_simt_exit): Ditto.
28407 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28409 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28410 (lower_rec_input_clauses): Likewise.
28411 (lower_lastprivate_clauses): Handle SIMT privatization.
28413 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28414 (ompdevlow_adjust_simt_enter): New.
28415 (find_simtpriv_var_op): New.
28416 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28417 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28419 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28420 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28421 (copy_decl_for_dup_finish): Ditto.
28423 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28425 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28428 * config/i386/i386.c (ix86_option_override_internal): Always
28429 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28431 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28433 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28435 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28437 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28438 mark new edge's irreducible flag accordign to it.
28439 (vect_do_peeling): Check loop preheader edge's irreducible flag
28440 and pass it to function slpeel_add_loop_guard.
28442 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28444 PR tree-optimization/80218
28445 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28446 Update block frequencies and counts.
28448 2017-03-28 Richard Biener <rguenther@suse.de>
28450 PR tree-optimization/78644
28451 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28452 of a simplification result we may not use it at all.
28454 2017-03-28 Richard Biener <rguenther@suse.de>
28457 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28458 without arguments, generate default definition of a SSA name.
28460 2017-03-28 Richard Biener <rguenther@suse.de>
28462 PR middle-end/80222
28463 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28464 TYPE_REF_CAN_ALIAS_ALL references.
28465 * fold-const.c (fold_indirect_ref_1): Likewise.
28467 2017-03-28 Martin Liska <mliska@suse.cz>
28470 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28471 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28473 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28474 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28476 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28477 (EXTRA_SPECS): Define.
28478 (SUBTARGET_EXTRA_SPECS): Likewise.
28479 (SUBTARGET_CPP_SPEC): Likewise.
28480 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28481 SUBTARGET_EXTRA_SPECS.
28482 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28484 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28486 * config/arc/simdext.md (vst64_insn): Update pattern.
28487 (vld32wh_insn): Likewise.
28488 (vld32wl_insn): Likewise.
28489 (vld64_insn): Likewise.
28490 (vld32_insn): Likewise.
28492 2017-03-28 Marek Polacek <polacek@redhat.com>
28495 * fold-const.c (fold_comparison): Use protected_set_expr_location
28496 instead of SET_EXPR_LOCATION.
28498 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28500 * tree.c (add_expr): Avoid name lookup warning.
28502 2017-03-27 Jeff Law <law@redhat.com>
28504 PR tree-optimization/80216
28505 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28506 function name. Limit recursion depth.
28507 (record_temporary_equivalences): Corresponding changes.
28509 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28511 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28514 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28517 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28519 * cfgcleanup.c (reg_note_cfa_p): New array.
28520 (insns_have_identical_cfa_notes): New function.
28521 (old_insns_match_p): Don't cross-jump in between /f
28522 and non-/f instructions. If both i1 and i2 are frame related,
28523 verify all CFA notes, their order and content.
28525 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28528 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28529 HImode and SImode with zero extend to DImode to one insn.
28530 (bswap<mode>2_extenddi): Likewise.
28531 (bswapsi2_extenddi): Likewise.
28532 (bswaphi2_extendsi): Likewise.
28533 (bswaphi2): Combine bswap HImode and SImode into one insn.
28534 Separate memory insns from swapping register.
28535 (bswapsi2): Likewise.
28536 (bswap<mode>2): Likewise.
28537 (bswaphi2_internal): Delete, no longer used.
28538 (bswapsi2_internal): Likewise.
28539 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28540 store, and gpr<-gpr swap insns.
28541 (bswap<mode>2_store): Likewise.
28542 (bswaphi2_reg): Register only splitter, combine with the splitter.
28543 (bswaphi2 splitter): Likewise.
28544 (bswapsi2_reg): Likewise.
28545 (bswapsi2 splitter): Likewise.
28546 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28547 the insns into load, store, and register/register insns.
28548 (bswapdi2_ldbrx): Likewise.
28549 (bswapdi2_load): Likewise.
28550 (bswapdi2_store): Likewise.
28551 (bswapdi2_reg): Likewise.
28553 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28555 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28556 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28558 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28561 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28563 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28564 special handling for target option conflicts between dform
28565 options (-mpower9-dform, -mpower9-dform-vector,
28566 -mpower9-dform-scalar) and -mno-direct-move.
28568 2017-03-27 Richard Biener <rguenther@suse.de>
28570 PR tree-optimization/80181
28571 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28573 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28575 * config/arc/predicates.md (move_double_src_operand): Replace the
28576 call to move_double_src_operand with a call to address_operand.
28578 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28580 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28581 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28582 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28584 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28586 * config/arc/predicates.md (long_immediate_loadstore_operand):
28587 Consider scaled addresses cases.
28589 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28591 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28592 restored when in interrupt.
28593 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28594 doesn't have delay slot.
28596 2017-03-27 Richard Biener <rguenther@suse.de>
28599 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28600 inlined thunk clones.
28602 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28605 * asan.c (instrument_derefs): Copy over last operand from
28606 original COMPONENT_REF to the new COMPONENT_REF with
28607 DECL_BIT_FIELD_REPRESENTATIVE.
28608 * ubsan.c (instrument_object_size): Likewise.
28610 2017-03-27 Richard Biener <rguenther@suse.de>
28612 PR tree-optimization/80170
28613 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28614 sure DR/SCEV didnt fold in constants we do not see when looking
28615 at the reference base alignment.
28617 2017-03-27 Richard Biener <rguenther@suse.de>
28619 PR middle-end/80171
28620 * gimple-fold.c (fold_ctor_reference): Properly guard against
28621 NULL return value from canonicalize_constructor_val.
28623 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28626 * config/i386/i386.c (ix86_expand_builtin)
28627 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28628 flags reg setting and flags reg using instructions.
28629 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28630 clobbering instructions to zero extend op2.
28632 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28634 * doc/install.texi (Configuration) <--with-aix-soname>:
28635 Update link to AIX ld.
28637 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28639 PR rtl-optimization/80160
28640 PR rtl-optimization/80159
28641 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28642 reg_alternate_class into account.
28644 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28647 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28648 to consider in curr_insn_transform.
28650 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28652 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28653 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28654 and emit_mode_inner.
28656 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28658 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28659 argument to the overloaded builtin variants. Use the new flag to
28660 deprecate certain builtin variants.
28661 * config/s390/s390-builtin-types.def: Add new builtin types.
28662 * config/s390/s390-builtins.h: Support new flags field for
28663 overloaded builtins.
28664 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28665 (s390_macro_to_expand): Enable vector float data type.
28666 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28667 builtins by incrementing the __VEC__ version number.
28668 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28670 (s390_resolve_overloaded_builtin): Emit error messages depending
28671 on the builtin flags.
28672 * config/s390/s390.c (s390_expand_builtin): Support additional
28673 flags argument. Change error message to match the messages
28674 emitted in s390-c.c.
28675 * config/s390/s390.md: New UNSPEC_* constants.
28676 (op_type): Add new instruction types.
28677 * config/s390/vecintrin.h: Add new builtins and test data class
28679 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28680 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28681 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28682 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28683 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28684 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28686 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28687 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28688 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28689 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28691 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28692 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28693 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28694 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28696 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28697 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28698 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28699 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28700 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28701 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28702 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28704 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28705 ("vec_scatter_element<V_HW_4:mode>_DI")
28706 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28707 ("vec_fpint<mode>", "vflls")
28708 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28709 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28710 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28711 ("*vec_cmphe<mode>_cc"): ... these.
28713 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28714 mode constant instead of magic value.
28716 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28718 * config/s390/s390.c (s390_expand_vec_compare): Support other
28719 vector floating point modes than just V2DF.
28720 (s390_expand_vcond): Likewise.
28721 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28722 (s390_cannot_change_mode_class): Prevent mode changes between TF
28723 and V1TF in vector registers.
28724 * config/s390/s390.md (DF, SF): New mode attributes.
28725 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28726 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28727 SFmode support for VRs.
28728 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28730 (VFT, VF_HW): New mode iterators.
28731 (vw, sdx): New mode attributes.
28732 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28733 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28734 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28735 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28736 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28737 also the new vector floating point modes. Renaming to ...
28739 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28740 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28741 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28742 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28743 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28744 ("vec_unordered<mode>"): ... these.
28746 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28747 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28748 ("*vec_extendv2df"): New insn definitions.
28750 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28752 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28753 ("mulditi3_2", "*muldi3_sign"): New patterns.
28754 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28755 rename the pattern definition.
28757 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28759 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28761 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28763 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28765 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28766 instruction if possible.
28767 * config/s390/vector.md (vec_halfnumelts): New mode
28769 ("*vec_vllezlf<mode>"): New pattern.
28771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28773 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28774 ("popcountv4si2", "popcountv2di2"): Rename to ...
28775 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28776 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28778 ("popcount<mode>2_vxe"): New pattern.
28780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28782 * common/config/s390/s390-common.c (processor_flags_table): Add
28784 * config.gcc: Add arch12.
28785 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28786 Default to arch12 for unknown CPU model numbers.
28787 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28788 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28789 PROCESSOR_max sanity check.
28790 * config/s390/s390-opts.h (enum processor_type): Add
28792 * config/s390/s390.c (processor_table): Add arch12.
28793 (s390_expand_builtin): Add check for B_VXE flag.
28794 (s390_issue_rate): Add PROCESSOR_ARCH12.
28795 (s390_get_sched_attrmask): Likewise.
28796 (s390_get_unit_mask): Likewise.
28797 (s390_sched_score): Enable z13 scheduling for arch12.
28798 (s390_sched_reorder): Likewise.
28799 (s390_sched_variable_issue): Likewise.
28800 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28802 (s390_tune_attr): Use z13 scheduling also for arch12.
28803 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28804 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28805 (TARGET_VXE_P): New macros.
28806 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28807 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28808 * config/s390/s390.opt: Add arch12 as processor_type.
28810 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28812 * config/s390/s390.md
28813 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28814 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28815 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28817 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28818 Rename expanders to ...
28820 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28821 ("fixuns_truncdddi2_emu"): ... these.
28823 ("fixuns_trunc<mode>si2_emu"): New expander.
28825 ("*fixuns_truncdfdi2_z13"): Rename to ...
28826 ("*fixuns_truncdfdi2_vx"): ... this.
28828 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28830 * config/s390/2964.md: Remove the single element vector compare
28831 instructions which are no longer used.
28832 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28834 (s390_canonicalize_comparison): Remove handling of DFmode
28836 (s390_expand_vec_compare_scalar): Remove function.
28837 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28838 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28840 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28844 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28846 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28847 will anyway by matched by mov<mode>_64dfp.
28849 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28851 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28852 vlef/vstef. Add missing operand to vleif.
28854 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28856 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28857 pair for all vector types with 64 bit elements.
28858 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28859 * config/s390/vector.md (V_HW_64): ... here.
28860 (V_128_NOSINGLE): New mode iterator.
28861 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28862 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28863 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28864 ("*vec_load_pairv2di"): Change to ...
28865 ("*vec_load_pair<mode>"): ... this one.
28867 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28869 * config/s390/constraints.md: Add comments.
28870 (jKK): Reject element sizes > 8 bytes.
28871 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28873 * config/s390/s390.md: Add the s_operand checks formerly in
28874 s390_split_ok_p to various splitters where they are still
28876 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28877 for 128 bit vectors. Plus two splitters.
28879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28881 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28884 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28887 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28888 error if the boundary argument is not constant.
28890 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28892 PR rtl-optimization/80112
28893 * loop-doloop.c (doloop_condition_get): Don't check condition
28894 if cmp isn't SET with IF_THEN_ELSE src.
28896 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28898 PR tree-optimization/80158
28899 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28900 replacing a candidate statement, also replace it for the
28901 candidate's alternate interpretation.
28902 (replace_rhs_if_not_dup): Likewise.
28903 (replace_one_candidate): Likewise.
28905 2017-03-24 Richard Biener <rguenther@suse.de>
28907 PR tree-optimization/80167
28908 * graphite-isl-ast-to-gimple.c
28909 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28911 (translate_isl_ast_to_gimple::get_rename): Likewise.
28913 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28915 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28916 handling of certain combinations of target options, including the
28917 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28918 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28920 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28923 * config/arm/arm.md (*load_multiple): Add reload_completed to
28924 matching condition.
28926 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28927 Richard Biener <rguenth@suse.de>
28929 PR tree-optimization/79908
28930 PR tree-optimization/80136
28931 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28932 been cast away, gimplify_and_add suffices.
28934 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28936 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28938 2017-03-23 Richard Biener <rguenther@suse.de>
28940 PR tree-optimization/80032
28941 * gimplify.c (gimple_push_cleanup): Forced unconditional
28942 cleanups still have to go to the conditional_cleanups
28945 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28947 PR tree-optimization/80072
28948 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28950 (next_operand_entry_id): Change type to unsigned int.
28951 (sort_by_operand_rank): Make sure to return the right return value
28952 even if unsigned fields are bigger than INT_MAX.
28953 (struct oecount): Change cnt and id type to unsigned int.
28954 (oecount_hasher::equal): Formatting fix.
28955 (oecount_cmp): Make sure to return the right return value
28956 even if unsigned fields are bigger than INT_MAX.
28957 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28960 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28961 TREE_READONLY on result if writing it more than once.
28964 * doc/invoke.texi (-fsanitize=thread): Document that with
28965 -fnon-call-exceptions atomics are not able to throw
28969 * tsan.c: Include tree-eh.h.
28970 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28971 maybe_clean_or_replace_eh_stmt where needed.
28972 (instrument_memory_accesses): Add cfg_changed argument.
28973 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28974 if it returned true.
28975 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28977 PR rtl-optimization/63191
28978 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28979 wrapper function, moved the whole old content into ...
28980 (ix86_delegitimize_address_1): ... this. New inline function.
28981 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28982 true as last argument instead of ix86_delegitimize_address.
28984 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28986 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28987 cortexa57_branch_cost.
28989 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28991 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28993 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28996 * doc/md.texi (Constraints): Document wA constraint.
28997 * config/rs6000/constraints.md (wA): New.
28998 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28999 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29000 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29001 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29003 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29006 * gimplify.c (is_oacc_declared): New function.
29007 (oacc_default_clause): Use it to set default flags for acc declared
29008 variables inside parallel regions.
29009 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29010 declared variables.
29011 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29012 declare attribute to any decl as necessary.
29014 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29017 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29018 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29019 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29020 (arm_arch_lpae): This.
29021 * config/arm/arm.c (arm_arch7ve): Rename into ...
29022 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29023 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29026 2017-03-22 Martin Liska <mliska@suse.cz>
29029 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29030 error message instead of an ICE.
29032 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29034 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29036 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29038 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29043 2017-03-21 Martin Sebor <msebor@redhat.com>
29045 * doc/extend.texi: Use "cannot" instead of "can't."
29046 * doc/hostconfig.texi: Same.
29047 * doc/install.texi: Same.
29048 * doc/invoke.texi: Same.
29049 * doc/loop.texi: Same.
29050 * doc/md.texi: Same.
29051 * doc/objc.texi: Same.
29052 * doc/rtl.texi: Same.
29053 * doc/tm.texi: Same.
29054 * doc/tm.texi.in: Same.
29055 * doc/trouble.texi: Same.
29057 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29060 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29061 (collect_checksum_attributes): Set it.
29062 (die_checksum_ordered): Use it.
29064 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29066 PR tree-optimization/79908
29067 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29068 change: For a VA_ARG whose LHS has been cast away, use
29069 force_gimple_operand to construct the side effects.
29071 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29073 PR translation/80001
29074 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29075 more amenable to translation.
29076 (oacc_loop_auto_partitions): Likewise.
29078 2017-03-21 Marek Polacek <polacek@redhat.com>
29079 Martin Sebor <msebor@redhat.com>
29081 PR tree-optimization/80109
29082 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29083 on INTEGRAL_TYPE_P.
29085 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29086 Segher Boessenkool <segher@kernel.crashing.org>
29089 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29090 check reg_used_between_p between insn and one of succ or succ2
29091 depending on if succ is artificial insn not inserted into insn
29094 2017-03-21 Martin Liska <mliska@suse.cz>
29096 PR gcov-profile/80081
29097 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29098 * doc/gcc.texi: Include gcov-dump stuff.
29099 * doc/gcov-dump.texi: New file.
29101 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29103 PR rtl-optimization/79150
29104 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29105 conditional jump, if the jump is the last insn of the loop.
29107 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29108 Richard Biener <rguenth@suse.de>
29110 PR tree-optimization/79908
29111 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29112 been cast away, use force_gimple_operand to construct the side
29115 2017-03-21 Martin Liska <mliska@suse.cz>
29117 PR libfortran/79956
29118 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29121 2017-03-21 Brad Spengler <spender@grsecurity.net>
29124 * plugin.c (htab_hash_plugin): New function.
29125 (add_new_plugin): Use it and adjust.
29126 (parse_plugin_arg_opt): Adjust.
29127 (init_one_plugin): Likewise.
29129 2017-03-21 Richard Biener <rguenther@suse.de>
29131 PR tree-optimization/80032
29132 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29133 if set force the cleanup to happen unconditionally.
29134 (gimplify_target_expr): Push inserted clobbers with force_uncond
29135 to avoid them being removed by control-dependent DCE.
29137 2017-03-21 Richard Biener <rguenther@suse.de>
29139 PR tree-optimization/80122
29140 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29141 va_arg_pack_len when the inlined call stmt requires pack
29143 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29145 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29148 * tsan.c (instrument_builtin_call): If the memory model argument
29149 is not a constant, assume it is valid.
29152 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29155 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29157 PR rtl-optimization/79910
29158 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29159 if its dest is used by an insn before I2 (other than the combined
29160 insns themselves, which are properly handled already).
29162 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29165 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29167 * combine.c (record_used_regs): New static function.
29168 (try_combine): Handle situations where there is an additional
29169 instruction between I2 and I3 which needs to have a LOG_LINK
29173 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29175 * combine.c (try_combine): Delete redundant i1 test. Call
29176 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29178 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29181 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29182 alternatives 13/14.
29184 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29186 PR tree-optimization/80054
29187 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29188 the optimization if a PHI or any of its arguments is not dominated
29189 by the candidate's basis. Use gphi* rather than gimple* as
29191 (replace_profitable_candidates): Clean up a gimple* variable that
29192 should be a gphi* variable.
29194 2017-03-20 Martin Sebor <msebor@redhat.com>
29197 * doc/extend.texi (attribute constructor): Document present limitation.
29199 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29202 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29203 __POWER9_VECTOR__ #ifdef control, change template definition to
29204 use Power9-specific built-in function.
29205 (vec_any_eq): Likewise.
29206 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29207 to control outcomes from this test.
29208 (vector_ae_<mode>p): For VEC_F modes, likewise.
29210 2017-03-20 Ian Lance Taylor <iant@google.com>
29212 * config/i386/i386.c (ix86_function_regparm): Save an extra
29213 register for -fsplit-stack with DECL_STATIC_CHAIN.
29215 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29218 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29219 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29221 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29223 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29225 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29228 2017-03-20 Marek Polacek <polacek@redhat.com>
29231 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29233 2017-03-20 Richard Biener <rguenther@suse.de>
29235 PR tree-optimization/80113
29236 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29237 allocate extra SSA name for PHI def.
29238 (add_close_phis_to_outer_loops): Likewise.
29239 (add_close_phis_to_merge_points): Likewise.
29240 (copy_loop_close_phi_args): Likewise.
29241 (copy_cond_phi_nodes): Likewise.
29243 2017-03-20 Martin Liska <mliska@suse.cz>
29245 PR middle-end/79753
29246 * tree-chkp.c (chkp_build_returned_bound): Do not build
29247 returned bounds for a LHS that's not a BOUNDED_P type.
29249 2017-03-20 Martin Liska <mliska@suse.cz>
29253 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29254 COMPLEX_CST and VECTOR_CST.
29256 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29259 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29260 target operand. A new splitter adds the clobber statement in case
29261 the target operand is dead anyway.
29263 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29265 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29266 to age-old versions of binutils and glibc.
29268 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29270 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29272 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29274 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29276 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29278 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29279 requirement for binutils 2.13.
29281 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29283 * combine.c (try_combine): Delete redundant i1 test. Call
29284 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29286 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29288 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29289 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29291 <riscv64-*-elf>: Re-arrange section
29292 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29293 <riscv32-*-linux>: Likewise.
29294 <riscv64-*-elf>: Likewise
29295 <riscv64-*-linux>: Likewise.
29297 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29300 * aarch64.opt(verbose-cost-dump): Fix typo.
29302 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29305 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29306 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29308 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29310 * reload.c (find_reloads): When reloading a nonoffsettable address,
29311 use RELOAD_OTHER for it and its address reloads.
29313 PR rtl-optimization/79910
29314 * combine.c (record_used_regs): New static function.
29315 (try_combine): Handle situations where there is an additional
29316 instruction between I2 and I3 which needs to have a LOG_LINK
29319 2017-03-17 Jeff Law <law@redhat.com>
29321 PR tree-optimization/71437
29322 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29323 conditional in the hash table first.
29324 (vrp_dom_walker::before_dom_children): Extract condition from
29325 ASSERT_EXPR. Record condition, its inverion and any implied
29326 conditions as well.
29328 2017-03-17 Marek Polacek <polacek@redhat.com>
29329 Markus Trippelsdorf <markus@trippelsdorf.de>
29331 PR tree-optimization/80079
29332 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29335 2017-03-17 Richard Biener <rguenther@suse.de>
29337 PR middle-end/80075
29338 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29339 Properly verify the LHS before the RHS possibly claims to be
29341 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29344 2017-03-17 Martin Jambor <mjambor@suse.cz>
29346 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29347 (List of -O2 options): Likewise.
29348 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29351 2017-03-17 Tom de Vries <tom@codesourcery.com>
29353 * gcov-dump.c (print_usage): Print bug_report_url.
29355 2017-03-17 Richard Biener <rguenther@suse.de>
29357 PR middle-end/80050
29358 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29359 (parser::peek): Likewise.
29361 2017-03-17 Richard Biener <rguenther@suse.de>
29363 PR tree-optimization/80048
29364 * sese.c (free_sese_info): Properly release rename_map and
29365 copied_bb_map elements.
29367 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29369 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29370 Add linked-list forward and backlinks. Insert on
29371 construction, remove on destruction.
29372 (class pass_store_merging): Add m_stores_head field.
29373 (pass_store_merging::terminate_and_process_all_chains):
29374 Iterate over m_stores_head list.
29375 (pass_store_merging::terminate_all_aliasing_chains):
29377 (pass_store_merging::execute): Check for debug stmts first.
29378 Push new chains onto the m_stores_head stack.
29380 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29383 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29384 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29385 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29387 2017-03-16 Jeff Law <law@redhat.com>
29389 PR tree-optimization/71437
29390 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29391 member function. Implementation moved into after_dom_children
29392 member function and into the threader's thread_outgoing_edges
29394 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29395 some code into new thread_outgoing_edges.
29396 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29397 definition. Simplify marker handling (do it here). Assume we always
29398 have the available expression and the const/copies tables.
29399 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29401 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29402 * tree-vrp.c (equiv_stack): No longer file scoped.
29403 (vrp_dom_walker): New class.
29404 (vrp_dom_walker::before_dom_children): New member function.
29405 (vrp_dom_walker::after_dom_children): Likewise.
29406 (identify_jump_threads): Setup domwalker. Use it rather than
29407 walking edges in a random order by hand. Simplify setup/finalization.
29408 (finalize_jump_threads): Remove.
29409 (vrp_finalize): Do not call identify_jump_threads here.
29410 (execute_vrp): Do it here instead and call thread_through_all_blocks
29413 PR tree-optimization/71437
29414 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29416 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29418 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29419 (dom_opt_dom_walker::thread_across_edge): Remove
29420 handle_dominating_asserts argument. All callers changed.
29421 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29422 changes. Remove calls to lhs_of_dominating_assert. Other
29423 uses of handle_dominating_asserts turn into unconditional code
29424 (simplify_control_stmt_condition_1): Likewise.
29425 (simplify_control_stmt_condition): Likewise.
29426 (thread_through_normal_block, thread_across_edge): Likewise.
29427 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29428 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29429 object if it is not an SSA_NAME.
29430 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29431 before calling into the VRP specific simplifiers.
29432 (identify_jump_threads): Remove handle_dominating_asserts
29435 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29438 * tree-diagnostic.c (default_tree_printer): No longer static.
29439 * tree-diagnostic.h (default_tree_printer): New prototype.
29441 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29443 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29444 Change ins into fmov.
29446 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29448 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29449 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29450 Use h_con constraint for operand 1.
29451 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29452 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29454 2017-03-15 Jeff Law <law@redhat.com>
29456 PR tree-optimization/71437
29457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29458 (record_temporary_equivalences): Use it.
29460 PR tree-optimization/71437
29461 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29462 tree-ssa-scopedtables.
29463 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29464 (record_conditions, record_cond, vuse_eq): Likewise.
29465 (record_edge_info): Adjust to API tweak of record_conditions.
29466 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29467 (record_temporary_equivalences, optimize_stmt): Likewise.
29468 (eliminate_redundant_computations): Likewise.
29469 (record_equivalences_from_stmt): Likewise.
29470 * tree-ssa-scopedtables.c: Include options.h and params.h.
29471 (vuse_eq): New function, moved from tree-ssa-dom.c
29472 (build_and_record_new_cond): Likewise.
29473 (record_conditions): Likewise. Accept vector of conditions rather
29474 than edge_equivalence structure for first argument.
29475 for the first argument.
29476 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29477 from tree-ssa-dom.c.
29478 (avail_exprs_stack::record_cond): Likewise.
29479 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29480 from tree-ssa-dom.c.
29481 (avail_exprs_stack): Add new member functions lookup_avail_expr
29483 (record_conditions): Declare.
29485 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29488 * lra-constraints.c (process_alt_operands): Increase reject for
29489 reloading an input/output operand.
29491 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29494 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29495 insns to convert from signed/unsigned char/short to IEEE 128-bit
29497 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29499 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29502 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29503 subreg of inner mode for values already in registers.
29505 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29507 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29508 iteration reg is used after the loop.
29510 2017-03-14 Martin Sebor <msebor@redhat.com>
29512 PR tree-optimization/79800
29513 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29514 precision in negative-positive range.
29515 (format_floating): Call non-const overload with adjusted precision.
29517 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29520 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29523 2017-03-14 Martin Sebor <msebor@redhat.com>
29525 PR middle-end/80020
29526 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29527 * builtins.def (aligned_alloc): Use it.
29530 * Makefile.in (GTFILES): Add calls.c.
29531 * calls.c: Include "gt-calls.h".
29533 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29535 PR rtl-optimization/79728
29536 * regs.h (struct target_regs): New field
29537 x_contains_allocatable_regs_of_mode.
29538 (contains_allocatable_regs_of_mode): New macro.
29539 * reginfo.c (init_reg_sets_1): Initialize it, and change
29540 contains_reg_of_mode so it includes global regs as well.
29541 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29542 rather than contains_regs_of_mode.
29544 2017-03-14 Martin Liska <mliska@suse.cz>
29546 * doc/invoke.texi: Document options that can't be combined with
29547 -fcheck-pointer-bounds.
29549 2017-03-14 Martin Liska <mliska@suse.cz>
29551 PR middle-end/79831
29552 * doc/invoke.texi (-Wchkp): Document the option.
29554 2017-03-14 Martin Liska <mliska@suse.cz>
29556 * Makefile.in: Install gcov-dump.
29558 2017-03-14 Martin Liska <mliska@suse.cz>
29560 * multiple_target.c (expand_target_clones): Bail out for
29561 an invalid attribute.
29563 2017-03-14 Richard Biener <rguenther@suse.de>
29565 * alias.c (struct alias_set_entry): Pack properly.
29566 * cfgloop.h (struct loop): Likewise.
29567 * cse.c (struct set): Likewise.
29568 * ipa-utils.c (struct searchc_env): Likewise.
29569 * loop-invariant.c (struct invariant): Likewise.
29570 * lra-remat.c (struct cand): Likewise.
29571 * recog.c (struct change_t): Likewise.
29572 * rtl.h (struct address_info): Likewise.
29573 * symbol-summary.h (function_summary): Likewise.
29574 * tree-loop-distribution.c (struct partition): Likewise.
29575 * tree-object-size.c (struct object_size_info): Likewise.
29576 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29577 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29578 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29579 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29580 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29581 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29582 (struct _stmt_vec_info): Likewise.
29584 2017-03-14 Martin Liska <mliska@suse.cz>
29587 * multiple_target.c (create_dispatcher_calls): Check that
29588 a target can create a function dispatcher.
29590 2017-03-14 Martin Liska <mliska@suse.cz>
29593 * multiple_target.c (expand_target_clones): Drop local.local
29594 flag for default implementation.
29596 2017-03-14 Richard Biener <rguenther@suse.de>
29598 PR tree-optimization/80030
29599 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29601 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29603 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29604 gcc_fallthrough() instead of __attribute__((fallthrough));
29606 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29608 * doc/gcc.texi: Remove "up" link to (DIR).
29609 * doc/gccint.texi: Ditto.
29611 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29613 * doc/install.texi (Specific) <avr>: Remove reference to
29616 2017-03-13 Jeff Law <law@redhat.com>
29618 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29619 attribute rather than comments.
29621 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29622 match_scratch operand is highest.
29624 2017-03-13 Martin Liska <mliska@suse.cz>
29626 PR middle-end/78339
29627 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29628 is a CHKP clone, use original declaration.
29630 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29632 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29633 (arc_conditional_register_usage): Use a different allocation order
29634 when optimizing for size.
29635 * common/config/arc/arc-common.c (arc_option_optimization_table):
29636 Section anchors default on when optimizing for size.
29638 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29640 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29642 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29644 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29645 * config/arc/arc.md (cpu_facility): Add cd variant.
29646 (*movqi_insn): Add code density variant.
29647 (*movhi_insn): Likewise.
29648 (*movqi_insn): Likewise.
29649 (*addsi3_mixed): Likewise.
29650 (subsi3_insn): Likewise.
29652 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29654 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29656 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29658 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29659 expressions with MINUS and UNARY ops.
29661 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29664 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29666 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29667 between vec_select and vector argument.
29668 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29669 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29670 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29671 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29672 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29673 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29675 2017-03-13 Richard Biener <rguenther@suse.de>
29678 * params.def (vect-max-peeling-for-alignment): Fix typo.
29680 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29682 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29683 issue that only occurred with binutils below 2.18.
29685 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29687 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29688 refer to binutils 2.11/2.12 minimum.
29690 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29692 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29693 ftp.kernel.org and simplify binutils requirement.
29695 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29697 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29699 (Optimize Options): Ditto. Also remove redundancy.
29701 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29703 PR translation/79848
29704 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29706 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29707 to G_ to avoid double translation.
29709 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29711 PR translation/79923
29712 * auto-profile.c (get_combined_location): Convert leading
29713 character of diagnostics to lower case and remove trailing period.
29714 (read_profile): Likewise for various diagnostics.
29715 * config/arm/arm.c (arm_option_override): Remove trailing period
29716 from various diagnostics.
29717 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29718 (msp430_expand_delay_cycles): Likewise.
29720 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29723 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29724 full command-line argument, rather than just "str".
29725 (aarch64_validate_march): Likewise.
29726 (aarch64_validate_mtune): Likewise.
29728 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29730 PR rtl-optimization/78911
29731 * lra-assigns.c (must_not_spill_p): New function.
29732 (spill_for): Use it.
29734 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29736 PR tree-optimization/79981
29737 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29738 ATOMIC_COMPARE_EXCHANGE ifn result.
29739 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29740 IFN_ATOMIC_COMPARE_EXCHANGE.
29742 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29745 * opts.c (parse_sanitizer_options): Add missing question mark to
29746 "did you mean" message.
29748 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29750 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29752 (VMULEUH_UNS): Likewise.
29753 (VMULOUB_UNS): Likewise.
29754 (VMULOUH_UNS): Likewise.
29755 * config/rs6000/rs6000.c (builtin_function_type): Remove
29756 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29758 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29761 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29762 x with result of extra_parsing_for_operand_code_0.
29763 (function_reader::extra_parsing_for_operand_code_0): Convert
29764 return type from void to rtx, returning x. When reading
29765 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29766 larger size containing struct block_symbol.
29768 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29770 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29771 -mfloat128-hardware without -m64.
29773 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29776 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29777 entries to the case statement that marks unsigned arguments to
29778 overloaded functions.
29780 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29782 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29783 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29785 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29788 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29789 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29791 2017-03-10 Martin Liska <mliska@suse.cz>
29795 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29796 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29799 2017-03-10 Olivier Hainque <hainque@adacore.com>
29801 * tree-switch-conversion (array_value_type): Start by resetting
29802 candidate type to it's main variant.
29804 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29806 PR rtl-optimization/79909
29807 * combine.c (try_combine): Use simplify_replace_rtx on individual
29808 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29809 of the whole CALL_INSN_FUNCTION_USAGE.
29811 PR tree-optimization/79972
29812 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29813 get_range_info on SSA_NAMEs. Formatting fixes.
29815 2017-03-10 Richard Biener <rguenther@suse.de>
29816 Jakub Jelinek <jakub@redhat.com>
29818 PR tree-optimization/77975
29819 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29820 edge to be constant.
29821 (get_val_for): For constant x return it. Formatting fix.
29822 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29823 would use the same bases as the current one.
29825 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29827 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29828 instead of vec_select for V1TImode.
29829 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29831 (VSX_LE_128): Add V1TI to this mode iterator.
29832 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29833 (*vsx_le_perm_store_<mode>): Likewise.
29834 (pre-reload splitter for VSX stores): Likewise.
29835 (post-reload splitter for VSX stores): Likewise.
29836 (*vsx_xxpermdi2_le_<mode>): Likewise.
29837 (*vsx_lxvd2x2_le_<mode>): Likewise.
29838 (*vsx_stxvd2x2_le_<mode>): Likewise.
29840 2017-03-09 Michael Eager <eager@eagercon.com>
29842 Correct failures with --enable-checking=yes,rtl.
29844 * config/microblaze/microblaze.c (microblaze_expand_shift):
29845 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29846 test for const0_rtx.
29847 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29848 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29850 2017-03-09 Richard Biener <rguenther@suse.de>
29852 PR tree-optimization/79977
29853 * graphite-scop-detection.c (scop_detection::merge_sese):
29854 Handle the case of extra exits to blocks dominating the entry.
29856 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29858 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29861 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29863 PR rtl-optimization/79949
29864 * lra-constraints.c (process_alt_operands): Check memory when
29865 trying to predict a cycle. Print about the overall increase.
29867 2017-03-09 Richard Biener <rguenther@suse.de>
29869 PR middle-end/79971
29870 * gimple-expr.c (useless_type_conversion_p): Preserve
29871 TYPE_SATURATING for fixed-point types.
29873 2017-03-09 Richard Biener <rguenther@suse.de>
29876 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29877 alignment of BLKmode params.
29879 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29882 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29883 (VALL_NO_V2Q): Likewise.
29885 * config/aarch64/aarch64-simd.md
29886 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29888 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29889 VALL_NO_V2Q mode iterator.
29890 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29892 2017-03-09 Martin Liska <mliska@suse.cz>
29894 PR tree-optimization/79631
29895 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29896 tree_int_cst_sign_bit just for INTEGER constants.
29898 2017-03-09 Martin Liska <mliska@suse.cz>
29902 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29905 2017-03-09 Marek Polacek <polacek@redhat.com>
29908 * tree.c (inchash::add_expr): Handle TREE_VEC.
29910 2017-03-09 Martin Liska <mliska@suse.cz>
29913 (chkp_narrow_size_and_offset): New function.
29914 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29915 (void chkp_parse_bit_field_ref): New function.
29916 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29917 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29919 2017-03-09 Martin Liska <mliska@suse.cz>
29922 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29923 (chkp_find_bounds_1): Remove gcc_unreachable.
29925 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29928 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29929 BUILT_IN_SYNC*, determine the access type from the size suffix and
29930 always build a MEM_REF with that type. Handle forgotten
29931 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29934 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29935 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29936 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29937 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29938 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29939 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29940 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29941 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29942 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29943 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29944 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29945 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29946 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29947 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29948 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29949 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29950 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29951 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29952 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29953 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29954 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29955 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29956 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29957 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29958 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29959 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29960 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29961 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29962 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29963 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29964 definitions outside of __OPTIMIZE__ guarded section.
29967 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29968 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29969 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29970 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29973 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29975 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29976 ("vfenez<mode>"): Add missing constraints.
29978 2017-03-08 Martin Sebor <msebor@redhat.com>
29981 * config/nds32/nds32.c (nds32_option_override):
29982 Fix misspelled diagnostic.
29984 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29987 * gimplify.c (gimplify_omp_for): Replace index var in outer
29988 taskloop statement with an artificial variable and add
29989 OMP_CLAUSE_PRIVATE clause for it.
29991 2017-03-08 Richard Biener <rguenther@suse.de>
29993 PR tree-optimization/79955
29994 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29995 for accesses that are completely outside of the variable.
29997 2017-03-08 Andrew Haley <aph@redhat.com>
29999 PR tree-optimization/79943
30000 * tree-ssa-loop-split.c (compute_new_first_bound): When
30001 calculating the new upper bound, (END-BEG) should be added, not
30004 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30006 * config/avr/avr.md (setmemhi): Make sure match_dup
30007 operand number comes before match_scratch.
30009 2017-03-08 Richard Biener <rguenther@suse.de>
30011 PR tree-optimization/79920
30012 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30013 with ncopies == 1 to ...
30014 (vect_transform_slp_perm_load): ... here. Properly compute
30015 all element loads by iterating VF times over the group. Do
30016 not handle ncopies (computed in a broken way) in
30017 vect_create_mask_and_perm.
30019 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30022 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30023 is a uniform vector, use uniform_vector_p return value instead of
30024 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30026 2017-03-07 Marek Polacek <polacek@redhat.com>
30028 PR middle-end/79809
30029 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30030 (alloca_call_type): Likewise.
30032 2017-03-07 Martin Liska <mliska@suse.cz>
30034 * gcov.c (process_args): Put comment to correct location.
30036 2017-03-07 Martin Liska <mliska@suse.cz>
30038 PR middle-end/68270
30039 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30040 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30041 (chkp_narrow_bounds_for_field): Likewise.
30042 (chkp_parse_array_and_component_ref): Pass one more argument to
30045 2017-03-07 Richard Biener <rguenther@suse.de>
30047 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30050 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30052 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30053 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30055 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30058 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30059 to end of description.
30060 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30062 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30064 PR rtl-optimization/79901
30065 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30067 (*avx512f_<code><mode>3<mask_name>): ... this.
30068 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30069 iterator instead of VI8_AVX2_AVX512BW.
30071 PR rtl-optimization/79901
30072 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30073 min/max expander, expand it using expand_vec_cond_expr.
30076 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30079 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30082 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30083 to void * for PCH reasons.
30084 * dwarf2out.c (output_loc_operands, output_die): Cast
30085 v.val_vec.array to unsigned char *.
30087 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30090 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30093 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30095 PR rtl-optimization/79571
30096 * lra-constraints.c (process_alt_operands): Calculate static
30097 reject and subtract it from overall when only addresses will be
30100 2017-03-06 Julia Koval <julia.koval@intel.com>
30103 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30104 incoming stack boundary to 128 for 64-bit targets.
30106 2017-03-06 Richard Biener <rguenther@suse.de>
30108 PR tree-optimization/79894
30109 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30110 to NULL after folding it.
30112 2017-03-06 Richard Biener <rguenther@suse.de>
30114 PR tree-optimization/79824
30115 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30116 check disabling peeling for gaps.
30118 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30120 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30121 attributes): Document gettimeofday.
30123 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30125 * config/s390/s390.c (s390_option_override_internal): Set
30126 PARAM_MIN_VECT_LOOP_BOUND
30128 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30130 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30131 * config/s390/s390.md: Likewise.
30133 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30136 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30137 (<avx2_avx512>_perm<mode>): Rename to ...
30138 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30140 (<avx512>_perm<mode>_mask): Rename to ...
30141 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30143 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30144 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30145 instead of VI8F_256_512.
30146 (avx512f_perm<mode>): New define_expand.
30147 (avx512f_perm<mode>_mask): Likewise.
30148 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30149 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30151 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30153 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30154 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30156 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30158 2017-03-06 Martin Liska <mliska@suse.cz>
30161 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30162 when having a SSA NAME w/o VAR_DECL assigned to it.
30164 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30166 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30167 msa_dpsub_<su>_d): Fix MODE for vec_select.
30169 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30171 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30173 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30175 2017-03-06 Richard Biener <rguenther@suse.de>
30177 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30178 * plugin.c (register_plugin_info): Likewise.
30179 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30181 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30183 * config/i386/sse.md (sse_storehps, sse_storelps,
30184 avx_<castmode><avxsizesuffix>_<castmode>,
30185 avx512f_<castmode><avxsizesuffix>_<castmode>,
30186 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30187 in condition that at least one operand is not a MEM.
30189 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30191 PR middle-end/79805
30192 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30193 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30195 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30196 gimple_call_nothrow_p flag based on whether original builtin can throw.
30197 If it can, emit following stmts on the fallthrough edge.
30198 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30199 don't create new bb if inserting just debug stmts on the edge, try to
30200 insert them on the fallthru bb or just reset debug stmts.
30202 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30205 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30206 restore recog_data (including the operand rtxes inside it) around
30207 the call to get_insn_template.
30209 2017-03-03 Martin Sebor <msebor@redhat.com>
30211 PR tree-optimization/79699
30212 * context.c (context::~context): Free MPFR caches to avoid
30213 a memory leak on program exit.
30215 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30217 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30218 Use wide_int::ulow () instead of .elt (0).
30220 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30222 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30223 (*pushxf): Limit oF constraint to 32bit targets and add oC
30224 constraint for 64bit targets.
30225 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30226 (*pushdf): Change rmF constraint to rmC.
30228 2017-03-03 Martin Liska <mliska@suse.cz>
30230 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30231 Remove unused variable.
30233 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30236 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30237 is a memory operand, increase num_memory.
30238 (ix86_expand_args_builtin): Likewise.
30240 2017-03-03 Jan Hubicka <jh@suse.cz>
30243 * ipa-devirt.c (maybe_record_node): Properly handle
30244 __cxa_pure_virtual visibility.
30246 2017-03-03 Martin Liska <mliska@suse.cz>
30248 PR tree-optimization/79803
30249 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30251 (pass_loop_prefetch::execute): Disabled optimization if an
30252 assumption about L1 cache size is not met.
30254 2017-03-03 Martin Liska <mliska@suse.cz>
30256 PR rtl-optimization/79574
30257 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30258 (hash_scan_set): Likewise.
30259 (dump_hash_table): Likewise.
30260 (hoist_code): Likewise.
30262 2017-03-03 Richard Biener <rguenther@suse.de>
30264 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30265 in place of uhigh ().
30266 (fixed_convert_from_real): Likewise.
30268 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30271 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30273 2017-03-03 Richard Biener <rguenther@suse.de>
30275 PR middle-end/79818
30276 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30277 TYPE_OVERFLOW_UNDEFINED check.
30279 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30281 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30283 (vector_ae_<mode>_p): Likewise.
30284 (vector_nez_<mode>_p): Likewise.
30285 (vector_ne_v2di_p): Likewise.
30286 (vector_ae_v2di_p): Likewise.
30287 (vector_ne_<mode>_p): Likewise.
30288 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30290 (vsx_tsqrt<mode>2_fe): Likewise.
30292 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30295 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30297 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30299 PR rtl-optimization/79780
30300 * cprop.c (one_cprop_pass): When second and further conditional trap
30301 in a single basic block is turned into an unconditional trap, turn it
30302 into a deleted note to avoid RTL verification failures.
30304 2017-03-02 Richard Biener <rguenther@suse.de>
30306 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30308 2017-03-02 Richard Biener <rguenther@suse.de>
30310 PR tree-optimization/79345
30312 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30313 param and abort the walk, returning -1 if it is hit.
30314 (walk_aliased_vdefs): Take a limit param and pass it on.
30315 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30316 defaulting to 0 and return a signed int.
30317 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30318 (check_defs): New helper.
30319 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30320 about uninitialized memory.
30321 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30322 bogus uninitialized warning.
30323 (fixed_convert_from_real): Likewise.
30325 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30327 PR tree-optimization/66768
30328 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30329 iv_use if base object can't be determined.
30331 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30333 PR tree-optimization/79345
30334 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30335 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30336 (get_pattern_stats): Initialize it.
30337 * genemit.c (gen_expand): Verify match_scratch numbers come after
30338 match_operand/match_dup numbers.
30339 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30340 match_scratch numbers.
30341 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30343 * config/s390/s390.md (trunctdsd2): Likewise.
30345 2017-03-02 Richard Biener <rguenther@suse.de>
30347 * wide-int.h (wide_int_storage::operator=): Implement in terms
30350 2017-03-02 Richard Biener <rguenther@suse.de>
30352 PR tree-optimization/79777
30353 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30354 the to insert expression to sth existing.
30356 2017-03-01 Martin Sebor <msebor@redhat.com>
30358 PR middle-end/79692
30359 * gimple-ssa-sprintf.c
30360 (directive::known_width_and_precision): New function.
30361 (format_integer): Use it.
30362 (get_mpfr_format_length): Consider the full range of precision
30363 when computing %g output with the # flag. Set the likely byte
30364 count to 3 rather than 1 when precision is indeterminate.
30365 (format_floating): Correct the lower bound of precision.
30367 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30369 * doc/invoke.texi: Document default code model for 64-bit Linux.
30371 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30374 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30375 udiv rather than div since input pattern is unsigned.
30377 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30379 * config/i386/i386.c (print_reg): Warn for values of
30380 unsupported size in integer register.
30382 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30385 * config/rs6000/predicates.md (current_file_function_operand): Do
30386 not allow self calls to be local if the function is replaceable.
30388 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30391 * config/rs6000/altivec.h (vec_ctz and others): Change the
30392 preprocessor macro that controls conditional compilation from
30393 _ARCH_PWR9 to __POWER9_VECTOR__.
30394 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30395 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30396 control (instead of _ARCH_PWR9 control) so that template
30397 definition uses power9-specific function.
30398 (vec_any_eq): Likewise.
30399 (vec_all_ne): Change macro definition to use a power9-specific
30400 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30401 _ARCH_PWR9 control).
30402 (vec_any_eq) Likewise.
30403 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30404 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30405 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30406 support for xvcmpnedp instruction.
30407 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30408 macro expansion so that Power9 implementation of vec_all_ne does
30409 not use the AltiVec predicate framework.
30410 (VCMPNEH_P): Likewise.
30411 (VCMPNEW_P): Likewise.
30412 (VCMPNED_P): Likewise.
30413 (VCMPNEFP_P): Likewise.
30414 (VCMPNEDP_P): Likewise.
30415 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30416 implementation of vec_any_eq to not use AltiVec predicate
30418 (VCMPAEH_P): Likewise.
30419 (VCMPAEW_P): Likewise.
30420 (VCMPAED_P): Likewise.
30421 (VCMPAEFP_P): Likewise.
30422 (VCMPAEDP_P): Likewise.
30423 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30424 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30425 not use the AltiVec predicate framework.
30426 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30427 of vec_any_eq to not use AltiVec predicate framework.
30428 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30429 support for predefined __POWER9_VECTOR__ macro to indicate that
30430 Power9 instruction selection is enabled.
30431 (altivec_overloaded_builtins): Remove extraneous
30432 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30433 function argument types RS6000_BTI_bool_V16QI and
30434 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30435 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30436 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30437 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30438 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30439 Power9 for implementations of vec_cmpne. Change the signature for
30440 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30441 (representing vec_all_ne) to remove the previously described first
30442 argument of type RS6000_BTI_INTSI, as this was an artifact of
30443 reliance on the AltiVec predicate framework, which is no longer
30444 used in the implementation of these functions. Add
30445 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30446 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30447 since, unlike the AltiVec predicate framework implementation, we
30448 do not share function descriptors between vec_alle and vec_anyeq.
30449 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30450 set of modes that receive special treatment even when
30451 TARGET_P9_VECTOR is true. The special treatment emits code that
30452 does not depend on Power9 instructions.
30453 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30454 define_expand to not rely on AltiVec predicate framework.
30455 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30457 (vector_ne_v2di_p): Change this define_expand to not rely on
30458 AltiVec predicate framework.
30459 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30461 (vector_ne_<mode>_p): Change this define_expand to not rely on
30462 AltiVec predicate framework.
30463 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30465 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30466 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30467 define_insn pattern.
30468 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30469 define_insn pattern because the xvcmpne<VSs>. instruction is not
30471 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30472 instruction is not supported.
30474 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30476 * config/nvptx/nvptx.c: Include intl.h.
30478 2017-03-01 Martin Jambor <mjambor@suse.cz>
30481 * ipa-prop.h (ipa_bits): Removed field known.
30482 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30483 to pointers. Adjusted their comments to warn about their sharing.
30484 (ipcp_transformation_summary): Change bits to a vector of pointers.
30485 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30486 (ipa_get_ipa_bits_for_value): Declare.
30487 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30488 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30489 (ipa_bits_hash_table): Likewise.
30490 (ipa_vr_ggc_hash_traits): Likewise.
30491 (ipa_vr_hash_table): Likewise.
30492 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30493 being pointers and vr_known being removed.
30494 (ipa_set_jf_unknown): Likewise.
30495 (ipa_get_ipa_bits_for_value): New function.
30496 (ipa_set_jfunc_bits): Likewise.
30497 (ipa_get_value_range): New overloaded functions.
30498 (ipa_set_jfunc_vr): Likewise.
30499 (ipa_compute_jump_functions_for_edge): Use the above functions to
30500 construct bits and vr parts of jump functions.
30501 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30502 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30504 (ipcp_grow_transformations_if_necessary): Also allocate
30505 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30507 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30508 them. Fix too long lines.
30509 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30510 vr_known being removed.
30511 (ipa_read_jump_function): Use new setter functions to construct bits
30512 and vr parts of jump functions or set them to NULL.
30513 (write_ipcp_transformation_info): Adjust for bits being pointers.
30514 (read_ipcp_transformation_info): Likewise.
30515 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30517 Include gt-ipa-prop.h.
30518 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30520 (ipcp_store_bits_results): Likewise.
30521 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30522 Do not write to existing jump functions but use a temporary instead.
30524 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30527 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30528 attempt to use its first operand as BIT_FIELD_REF base.
30530 2017-03-01 Richard Biener <rguenther@suse.de>
30532 PR middle-end/79721
30533 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30534 interpolating formula in wrapping arithmetic.
30535 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30537 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30539 PR tree-optimization/79734
30540 * tree-vect-generic.c (expand_vector_condition): Optimize
30541 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30542 Handle VEC_COND_EXPR where comparison has different inner width from
30543 type's inner width.
30545 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30547 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30548 markup, and similar issues. Remove @opindex entries for things
30549 that aren't options. Add missing -mmpy-option entries.
30551 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30553 PR tree-optimization/79737
30554 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30555 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30556 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30557 instead of byte_size. Formatting fix.
30558 (shift_bytes_in_array_right): Formatting fix.
30560 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30563 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30564 condition on optimize for the leaf function test.
30566 2017-02-28 Martin Liska <mliska@suse.cz>
30569 * read-rtl-function.c (function_reader::handle_unknown_directive):
30570 Bail out when one uses -flto.
30572 2017-02-28 Martin Liska <mliska@suse.cz>
30574 * common.opt: Replace space with tabular for options of <number>
30576 * config/i386/i386.opt: Show <number> value for
30577 -mlarge-data-threshold.
30578 * opts.c (print_filtered_help): Do not display number in hexadecimal
30581 2017-02-28 Martin Liska <mliska@suse.cz>
30583 * common.opt: Fix --help=option -Q for options which are of
30586 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30588 * config/i386/i386.c (print_reg): Error out for values
30589 of 8-bit size in invalid integer register.
30591 2017-02-28 Martin Sebor <msebor@redhat.com>
30593 PR tree-optimization/79691
30594 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30596 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30599 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30600 gcc_unreachable with output_operand_lossage.
30602 2017-02-28 Richard Biener <rguenther@suse.de>
30604 PR tree-optimization/79740
30605 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30607 (visit_nary_op): Insert the nary into the hashtable if we
30608 pattern-matched sth.
30609 * tree-ssa-pre.c (eliminate_insert): Robustify.
30611 2017-02-28 Richard Biener <rguenther@suse.de>
30613 PR middle-end/79731
30614 * fold-const.c (decode_field_reference): Reject out-of-bound
30617 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30619 * config/i386/i386.c: Include intl.h.
30620 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30621 instead of just cond ? "..." : "...".
30622 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30623 * coverage.c (read_counts_file): Likewise.
30624 * omp-offload.c: Include intl.h.
30625 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30626 of just cond ? "..." : "...".
30627 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30628 of just cond ? "..." : "...".
30630 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30633 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30635 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30636 'tune for' CPU name.
30637 * config/arm/arm-cpu-data.h: Regenerated.
30639 2017-02-28 Richard Biener <rguenther@suse.de>
30641 PR tree-optimization/79732
30642 * tree-inline.c (expand_call_inline): Do not shadow var.
30644 2017-02-28 Richard Biener <rguenther@suse.de>
30646 PR tree-optimization/79723
30647 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30648 address-space properly.
30650 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30652 * doc/optinfo.texi (Optimization groups): Fix option used for
30654 * doc/invoke.texi (-fopt-info): Document "omp".
30655 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30656 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30657 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30658 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30659 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30661 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30663 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30664 OPTGROUP_OMP with "omp".
30666 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30669 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30670 for arithmetic shift of unsigned V2DI.
30672 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30674 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30675 arc/linux.h headers.
30676 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30677 (LINK_SPEC): Likewise.
30678 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30679 (EXTRA_SPECS): Likewise.
30680 (STARTFILE_SPEC): Likewise.
30681 (ENDFILE_SPEC): Likewise.
30682 (LIB_SPEC): Likewise.
30683 (TARGET_SDATA_DEFAULT): Likewise.
30684 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30685 (MULTILIB_DEFAULTS): Likewise.
30686 (DWARF2_UNWIND_INFO): Likewise.
30687 * config/arc/big.h: New file.
30688 * config/arc/elf.h: Likewise.
30689 * config/arc/linux.h: Likewise.
30690 * config/arc/t-uClibc: Remove.
30692 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30694 PR tree-optimization/77536
30695 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30696 (tree_transform_and_unroll_loop): Use above function to compute the
30697 estimated niter of unrolled loop and use it when scaling profile.
30698 Also use count info rather than frequency if it's non-zero.
30699 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30700 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30701 (vect_transform_loop): Call above function.
30703 2017-02-27 Richard Biener <rguenther@suse.de>
30705 PR tree-optimization/45397
30706 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30707 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30708 (visit_nary_op): Add pattern matching for CSEing sign-changed
30709 or truncated operations with wider ones.
30711 2017-02-27 Richard Biener <rguenther@suse.de>
30713 PR tree-optimization/79690
30714 * tree-vect-stmts.c (vectorizable_store): Use vector type
30715 built from the DR with address-space.
30717 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30719 * doc/invoke.texi (Optimize Options): Refine the description
30720 of asan-use-after-return.
30722 2017-02-25 Alan Modra <amodra@gmail.com>
30724 PR rtl-optimization/79584
30725 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30726 base, not ad->base_term, the reg within base. Remove assertion
30727 that ad->base == ad->base_term. Replace gen_int_mode using
30728 bogus mode with const0_rtx.
30730 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30732 PR middle-end/79396
30733 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30734 FMA_EXPR like tcc_binary or tcc_unary.
30736 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30739 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30741 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30742 (output_loc_operands): Handle DW_OP_call_ref and
30743 DW_OP_GNU_variable_value.
30744 (struct variable_value_struct): New type.
30745 (struct variable_value_hasher): Likewise.
30746 (variable_value_hash): New variable.
30747 (string_types): Remove.
30748 (copy_loc_descr): New function.
30749 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30750 (prepend_loc_descr_to_each): New function.
30751 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30752 instead of add_loc_descr_to_each if the first argument is single
30753 location list and the second has multiple.
30754 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30755 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30756 when looking for variable value which doesn't have other location info.
30757 (loc_list_from_tree): Formatting fix.
30758 (gen_array_type_die): Simplify DW_AT_string_length handling.
30759 (adjust_string_types): Remove.
30760 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30761 string_types. Call resolve_variable_values.
30762 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30763 (resolve_addr_in_expr): Likewise. Add A argument.
30764 (copy_deref_exprloc): Remove deref argument. Adjust for the
30765 original expression being DW_OP_GNU_variable_value with optionally
30766 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30767 optionally after it.
30768 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30769 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30770 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30771 (variable_value_hasher::hash, variable_value_hasher::equal): New
30773 (resolve_variable_value_in_expr, resolve_variable_value,
30774 resolve_variable_values, note_variable_value_in_expr,
30775 note_variable_value): New functions.
30776 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30779 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30782 * opts.h (handle_generated_option): Add GENERATED_P argument.
30783 * opts-common.c (handle_option): Adjust function comment.
30784 (handle_generated_option): Add GENERATED_P argument, pass it to
30786 (control_warning_option): Pass false to handle_generated_option
30788 * opts.c (maybe_default_option): Pass true to handle_generated_option
30790 * optc-gen.awk: Likewise.
30792 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30794 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30795 a REG, look at the REG it is a SUBREG of.
30796 (splitter for cmpeqsi_t): Ditto.
30798 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30800 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30801 the special USEs with the pattern of the insn, not the insn itself.
30803 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30806 * doc/invoke.texi: Document -mload-store-pairs.
30808 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30809 Sandra Loosemore <sandra@codesourcery.com>
30811 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30812 argument isn't a CONST_INT.
30813 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30814 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30815 (nios2_validate_compare): Bypass alternate compare logic if *op2
30816 is not a CONST_INT.
30817 (ldstwm_operation_p): Return false if first_base is not a REG or
30818 if first_offset is not a CONST_INT.
30820 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30822 * config/cris/cris.md: Use correct operand in a define_peephole2.
30824 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30826 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30828 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30830 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30831 this_insn if it is an INSN or JUMP_INSN.
30832 (force_offsettable): Look at base, not at addr.
30833 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30834 on things that aren't necessarily CONST_INTs.
30836 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30838 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30839 -mfpmath=sse is the default also for x86-32 targets with SSE2
30840 instruction set when @option{-ffast-math} is enabled
30842 2017-02-24 Jeff Law <law@redhat.com>
30844 PR rtl-optimizatoin/79286
30845 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30848 2017-02-24 Richard Biener <rguenther@suse.de>
30850 PR tree-optimization/79389
30851 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30854 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30856 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30857 function comment to reflect reality.
30858 (loop_exits_before_overflow): Fix typo in function description.
30860 2017-02-24 Richard Biener <rguenther@suse.de>
30862 PR tree-optimization/79389
30863 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30864 properly that a threading opportunity exists. Detect conditional
30865 copy/constant propagation opportunities.
30867 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30869 * config/visium/visium.md (type): Add trap.
30870 (b): New mode attribute.
30871 (*btst): Rename into...
30872 (*btst<mode>): ...this and adjust.
30873 (*cbranchsi4_btst_insn): Rename into...
30874 (*cbranch<mode>4_btst_insn): ...this and adjust.
30875 (trap): New define_insn.
30877 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30879 PR tree-optimization/79389
30880 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30881 (noce_reversed_cond_code): New function.
30882 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30883 reversed_comparison_code. Formatting fix.
30884 (noce_try_store_flag): Test rev_cond != NULL in addition to
30885 reversed_comparison_code.
30886 (noce_try_store_flag_constants): Likewise.
30887 (noce_try_store_flag_mask): Likewise.
30888 (noce_try_addcc): Use rev_cond if non-NULL instead of
30889 reversed_comparison_code.
30890 (noce_try_cmove_arith): Likewise. Formatting fixes.
30891 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30892 (noce_find_if_block): Initialize rev_cond.
30893 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30894 instead of false as last argument never attempt to reverse it
30897 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30899 PR tree-optimization/79663
30900 * tree-predcom.c (combine_chains): Process refs in reverse order
30901 only for ZERO length chains, and add explaining comment.
30903 2017-02-23 Jeff Law <law@redhat.com>
30905 PR tree-optimization/79578
30906 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30907 in call to operand_equal_p.
30909 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30912 * config/i386/cpuid.h: Fix another undefined behavior.
30914 2017-02-23 Richard Biener <rguenther@suse.de>
30916 PR tree-optimization/79683
30917 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30918 vector types for data-refs.
30920 2017-02-23 Martin Liska <mliska@suse.cz>
30922 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30924 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30926 PR middle-end/79665
30927 * internal-fn.c (get_range_pos_neg): Moved to ...
30928 * tree.c (get_range_pos_neg): ... here. No longer static.
30929 * tree.h (get_range_pos_neg): New prototype.
30930 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30931 are known to be in between 0 and signed maximum inclusive, try to
30932 expand both unsigned and signed divmod and use the cheaper one from
30935 2017-02-22 Jeff Law <law@redhat.com>
30937 PR tree-optimization/79578
30938 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30939 to compare base operands.
30941 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30944 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30945 gpc_reg_operand instead of fpr_reg_operand.
30947 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30949 * config/mips/mips.c (mips_return_in_memory): Force FP
30950 vector types to be returned in memory for o32 ABI.
30952 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30954 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30955 instead of DW_TAG_member for static data member declarations and don't
30956 set no_linkage_name for static inline data members.
30957 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30960 2017-02-22 Martin Liska <mliska@suse.cz>
30962 * doc/invoke.texi: Replace inequality signs with square brackets
30965 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30968 * lra-constraints.c (simplify_operand_subreg): Handle
30969 WORD_REGISTER_OPERATIONS targets.
30971 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30974 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30975 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30976 elimination by swapping fld*.
30978 2017-02-22 Richard Biener <rguenther@suse.de>
30980 PR tree-optimization/79673
30981 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30982 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30983 irrelevant address-space qualifiers and avoiding a
30984 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30986 2017-02-22 Richard Biener <rguenther@suse.de>
30988 PR tree-optimization/79666
30989 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30990 to not symbolically negate if that may introduce undefined
30993 2017-02-22 Martin Liska <mliska@suse.cz>
30996 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30997 * data-streamer-out.c (streamer_write_gcov_count_stream):
30999 * value-prof.c (stream_out_histogram_value): Make assert more
31000 precise based on type of counter.
31002 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31005 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31006 Use nonimmediate_operand instead of memory_operand for operand 1.
31007 (float-extend standard_x87sse_constant_load splitter): Ditto.
31009 2017-02-21 Jeff Law <law@redhat.com>
31011 PR tree-optimization/79621
31012 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31013 blocks with edges to themselves.
31015 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31018 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31019 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31020 Use gimple_call_builtin_p.
31023 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31024 on temporarily removed DEBUG_INSNs.
31026 PR tree-optimization/79649
31027 * tree-loop-distribution.c (classify_partition): Give up on
31028 non-generic address space loads/stores.
31030 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31032 * doc/loop.texi (Loop manipulation): Remove nonexistent
31033 tree_ssa_loop_version from the documentation.
31034 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31036 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31039 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31040 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31041 * config/rs6000/rs6000.c: Include except.h.
31042 (rs6000_expand_split_stack_prologue): Call
31043 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31045 2017-02-21 Martin Jambor <mjambor@suse.cz>
31048 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31049 have been analyzed.
31051 2017-02-21 Martin Jambor <mjambor@suse.cz>
31053 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31054 for backward compatibility only.
31055 * doc/invoke.texi (Option Summary): Remove all references to
31056 -fipa-cp-alignment.
31058 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31062 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31064 * lra-constraints.c (curr_insn_transform): Handle
31065 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31067 2017-02-21 Martin Liska <mliska@suse.cz>
31069 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31071 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31073 PR translation/79638
31074 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31076 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31079 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31080 (arm_function_ok_for_sibcall): Return false for an indirect call by
31081 descriptor if all the argument registers are used.
31082 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31083 alignment of the function.
31085 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31087 PR tree-optimization/61441
31088 * simplify-rtx.c (simplify_const_unary_operation): For
31089 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31090 the sNaN unmodified.
31092 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31094 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31095 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31096 instead of SYSTEM_HEADER_DIR.
31098 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31099 Martin Liška <mliska@suse.cz>
31101 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31102 Fix typos and grammar, use active voice, and clarify.
31104 2017-02-20 Marek Polacek <polacek@redhat.com>
31106 PR middle-end/79537
31107 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31110 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31112 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31115 * config/i386/i386.c (ix86_expand_builtin): Handle
31116 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31117 ix86_builtins_isa[fcode].isa as a requirement of those
31118 flags and any other flag in the bitmask.
31119 (ix86_init_mmx_sse_builtins): Use 0 instead of
31120 ~OPTION_MASK_ISA_64BIT as mask.
31121 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31122 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31123 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31124 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31126 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31129 * lra-constraints.c (split_reg): Check requested split mode
31130 is supported by the register.
31132 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31134 * lra-constraints.c (simplify_operand_subreg): Remove early
31137 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31140 * lra-constraints.c (curr_insn_transform): Tighten condition
31141 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31143 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31146 * lra-constraints.c (curr_insn_transform): Handle
31147 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31149 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31152 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31154 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31156 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31159 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31162 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31164 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31167 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31169 * final.c (last_columnnum, override_columnnum): New variables.
31170 (final_start_function): Set last_columnnum, pass it to begin_prologue
31171 hook and pass 0 to dwarf2out_begin_prologue.
31172 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31173 to source_line debug hook.
31174 (notice_source_line): Compute last_columnnum and for debug_column_info
31175 return true on column changes.
31176 * debug.h (struct gcc_debug_hooks): Add column argument to
31177 source_line and begin_prologue hooks.
31178 (debug_nothing_int_charstar_int_bool): Remove prototype.
31179 (debug_nothing_int_int_charstar,
31180 debug_nothing_int_int_charstar_int_bool): New prototypes.
31181 (dwarf2out_begin_prologue): Add column argument.
31182 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31183 begin_prologue hooks.
31184 (debug_nothing_int_charstar_int_bool): Remove.
31185 (debug_nothing_int_int_charstar,
31186 debug_nothing_int_int_charstar_int_bool): New functions.
31187 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31188 through to dwarf2out_source_line.
31189 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31190 (dwarf2out_source_line): Add column argument, emit it if requested.
31191 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31193 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31194 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31195 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31196 through to dwarf2out_begin_prologue.
31197 (vmsdbgout_source_line): Add column argument, pass it through to
31198 dwarf2out_source_line.
31199 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31200 dbxout_source_line caller.
31201 (dbxout_source_line): Add column argument.
31203 * common.opt (gno-column-info, gcolumn-info): New options.
31204 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31205 (check_die): Also test for multiple DW_AT_decl_column attributes.
31206 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31207 DW_AT_decl_column if requested.
31208 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31210 (gen_variable_die): Likewise.
31211 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31212 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31215 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31216 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31217 (ix86_handle_option): Handle OPT_m3dnowa.
31218 * doc/invoke.texi (-m3dnowa): Document.
31219 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31220 -m3dnowa instead of -m3dnow -march=athlon.
31223 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31224 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31226 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31229 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31230 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31231 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31232 generator for vsx_xxpermdi_<mode>_be.
31233 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31234 force big-endian semantics.
31235 (vsx_xxpermdi_<mode>_be): New define_expand with same
31236 implementation as previous version of vsx_xxpermdi_<mode>.
31238 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31240 PR tree-optimization/79327
31241 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31242 variable, its initialization and use.
31244 2017-02-17 Julia Koval <julia.koval@intel.com>
31246 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31247 (OPTION_MASK_ISA_PKU_UNSET): New.
31248 (ix86_handle_option): Handle -mrdpid.
31249 * config/i386/cpuid.h (bit_RDPID): New.
31250 * config/i386/driver-i386.c (host_detect_local_cpu):
31251 Detect RDPID feature.
31252 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31253 * config/i386/i386-c.c (ix86_target_macros_internal):
31255 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31256 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31257 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31258 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31259 * config/i386/i386.md (define_insn "rdpid"): New.
31260 * config/i386/i386.opt Add -mrdpid.
31261 * config/i386/immintrin.h (_rdpid_u32): New.
31263 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31265 PR rtl-optimization/79541
31266 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31267 instead of transforming it into USE.
31269 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31271 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31272 If HONOR_SNANS (SFmode) force the input to a register.
31273 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31274 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31275 an frsp or similar insn.
31277 2017-02-17 Martin Liska <mliska@suse.cz>
31279 PR rtl-optimization/79577
31280 * params.def (selsched-max-sched-times): Increase minimum to 1.
31282 2017-02-17 Martin Liska <mliska@suse.cz>
31284 PR rtl-optimization/79574
31285 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31287 2017-02-17 Martin Liska <mliska@suse.cz>
31289 PR tree-optimization/79529
31290 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31291 ssa_defined_default_def_p to handle cases which are implicitly
31293 * tree-ssa.c (ssa_defined_default_def_p): New function.
31294 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31295 which are implicitly defined.
31296 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31298 2017-02-17 Richard Biener <rguenther@suse.de>
31300 PR middle-end/79576
31301 * params.def (max-ssa-name-query-depth): Limit to 10.
31303 2017-02-17 Richard Biener <rguenther@suse.de>
31305 PR tree-optimization/79552
31306 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31309 2017-02-17 Richard Biener <rguenther@suse.de>
31312 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31314 2017-02-17 Marek Polacek <polacek@redhat.com>
31316 PR middle-end/79536
31317 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31318 (fold_negate_expr): New wrapper.
31320 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31322 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31323 Correct terminology and de-emphasize pre-standard behavior.
31325 2017-02-16 Alan Modra <amodra@gmail.com>
31327 PR rtl-optimization/79286
31328 * ira.c (def_dominates_uses): New function.
31329 (update_equiv_regs): Don't create an equivalence for insns that
31330 may trap where the register def does not dominate the use.
31332 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31334 PR rtl-optimization/78127
31335 * lra.c (lra): Call lra_eliminate before finish the loop after
31338 2017-02-16 Richard Biener <rguenther@suse.de>
31340 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31342 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31343 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31344 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31345 (isl_val_int_from_wi): New function.
31346 (extract_affine_gmp): Rename to ...
31347 (extract_affine_wi): ... this, take a widest_int.
31348 (extract_affine_int): Just wrap extract_affine_wi.
31349 (add_param_constraints): Use isl_val_int_from_wi.
31350 (add_loop_constraints): Likewise, and extract_affine_wi.
31352 2017-02-15 Jeff Law <law@redhat.com>
31354 PR middle-end/79521
31355 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31356 ira_init_register_move_cost_if_necessary.
31358 2017-02-15 Martin Sebor <msebor@redhat.com>
31360 PR middle-end/32003
31361 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31362 removed in a prior commit.
31364 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31366 PR tree-optimization/79347
31367 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31368 counters during peeling.
31370 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31372 * Makefile.in (site.exp): Remove "set ISLVER".
31374 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31377 * real.c (real_from_integer): Call real_convert even for decimal.
31379 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31382 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31384 2017-02-14 Andrew Pinski <apinski@cavium.com>
31386 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31387 cores and change the partno/implementer to be correct.
31388 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31389 the 'B" as the implementer.
31390 * config/aarch64/aarch64-tune.md: Regenerate.
31392 2017-02-14 Carl Love <cel@us.ibm.com>
31394 * config/rs6000/rs6000.c: Add case statement entry to make the
31395 xvcvuxdsp built-in argument unsigned.
31396 * config/rs6000/vsx.md: Fix the source and return operand types so they
31397 match the instruction definitions from the ISA document. Fix typo
31398 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31401 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31404 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31405 member early_clobber_alts.
31406 * lra-lives.c (reg_early_clobber_p): New.
31407 (process_bb_lives): Use it.
31408 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31409 (debug_operand_data): Initialize early_clobber_alts.
31410 (setup_operand_alternative): Set up early_clobber_alts.
31411 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31412 alternatives to new_insn_reg.
31413 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31415 (lra_update_insn_regno_info): Pass the new arg.
31417 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31419 PR middle-end/79505
31420 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31421 (new_oacc_loop_raw): Don't clear already cleared fields.
31424 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31425 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31426 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31428 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31431 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31433 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31436 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31437 the extra instruction to the right place to store 128-bit constant
31440 2017-02-14 Martin Sebor <msebor@redhat.com>
31442 PR middle-end/79448
31443 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31444 warning for strings of unknown length.
31446 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31448 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31450 2017-02-14 Jeff Law <law@redhat.com>
31453 * ira-costs.c (scan_one_insn): Initialize register move costs
31454 for pseudos seen in USE/CLOBBER insns.
31456 PR tree-optimization/79095
31457 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31458 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31459 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31460 if the operands are known to be not equal, then the resulting range
31462 (intersect_ranges): If the new range is ~[0,0] and the old range is
31463 wide, then prefer ~[0,0].
31464 * tree-vrp.c (overflow_comparison_p_1): New function.
31465 (overflow_comparison_p): New function.
31466 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31467 if NAME is used in an overflow test.
31468 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31469 overflow check that can be expressed as an equality test, then adjust
31470 ops to be that equality test.
31472 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31474 * config/s390/s390-builtin-types.def: Remove flags argument.
31475 * config/s390/s390.c (s390_init_builtins): Likewise.
31477 2017-02-14 Martin Liska <mliska@suse.cz>
31479 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31480 vector. Fix trailing white spaces.
31482 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31484 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31487 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31489 PR rtl-optimization/68664
31490 * config/arm/arm.c (arm_sched_can_speculate_insn):
31491 New function. Declare prototype.
31492 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31494 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31496 PR rtl-optimization/68664
31497 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31499 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31501 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31503 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31504 max skip bytes for function, loop and jump.
31506 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31508 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31509 ABS_EXPR for gimple dump.
31511 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31514 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31516 PR tree-optimization/79408
31517 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31518 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31519 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31520 also if rhs1 is INTEGER_CST.
31522 2017-02-14 Richard Biener <rguenther@suse.de>
31524 PR middle-end/79432
31525 * tree-into-ssa.c (insert_phi_nodes): When the function can
31526 have abnormal edges rewrite SSA names with broken use-def
31527 dominance out of SSA and register them for PHI insertion.
31529 2017-02-13 Martin Sebor <msebor@redhat.com>
31531 PR middle-end/79496
31532 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31533 clearing info.nowrite flag when snprintf size argument is a range.
31535 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31537 * cprop.c (cprop_jump): Add missing space in string literal.
31538 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31539 (get_constraint_for_component_ref): Likewise.
31540 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31541 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31542 * lra-constraints.c (process_alt_operands): Likewise.
31543 * ipa-inline.c (inline_small_functions): Likewise.
31544 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31545 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31546 * trans-mem.c (diagnose_tm_1_op): Likewise.
31547 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31548 (grid_parallel_clauses_gridifiable): Likewise.
31550 * config/nvptx/mkoffload.c (process): Add space in between
31553 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31554 "MOD4_SSE_REGS" and "ALL_REGS".
31556 * spellcheck.c (test_data): Add , in between "foo" and "food".
31558 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31561 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31562 boundary crossing check and subsequent code generation agree.
31564 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31566 * config/aarch64/aarch64.c (has_memory_op): Delete.
31567 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31570 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31572 PR rtl-optimization/79388
31573 PR rtl-optimization/79450
31574 * combine.c (distribute_notes): When removing TEM_INSN for which
31575 corresponding dest has last value recorded, invalidate that last
31578 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31580 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31581 of explicit '@'. Add missing assembly comment marker on branch costs
31584 2017-02-13 Nathan Sidwell <nathan@acm.org>
31586 * gengtype-lex.l (<in_struct>): Add '/'.
31588 2017-02-13 Martin Liska <mliska@suse.cz>
31591 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31593 2017-02-13 Richard Biener <rguenther@suse.de>
31595 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31597 * configure: Re-generate.
31598 * config.in: Likewise.
31599 * graphite-dependences.c: Simplify as if
31600 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31601 * graphite-isl-ast-to-gimple.c: Likewise.
31602 * graphite-optimize-isl.c: Likewise.
31603 * graphite-poly.c: Likewise.
31604 * graphite-sese-to-poly.c: Likewise.
31605 * graphite.h: Likewise.
31606 * toplev.c: Include isl/version.h and use isl_version () for
31607 printing the ISL version.
31608 * doc/install.texi: Update ISL requirement.
31610 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31612 * doc/standards.texi (Standards): Update reference to
31615 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31617 * doc/extend.texi (Named Address Spaces): sourceware.org now
31619 * doc/install.texi (Binaries): Ditto.
31622 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31624 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31625 standard terminology "stringize"/"stringizing" throughout.
31626 * doc/cppinternals.texi: Likewise.
31628 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31630 * doc/extend.texi: Fix some spelling mistakes and typos.
31631 * doc/invoke.texi: Likewise.
31633 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31636 * params.def (inline-min-speedup) Change from 10 to 8.
31638 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31640 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31643 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31646 * ipa-inline-analysis.c (get_minimal_bb): New function.
31647 (record_modified): Use it.
31648 (remap_edge_change_prob): Handle also ancestor functions.
31650 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31652 * doc/contrib.texi (Contributors): Remove broken link into
31653 the Mauve CVS repository.
31655 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31657 PR middle-end/79454
31658 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31659 result computation whenever lhs doesn't have vector mode, not
31660 just when it has BLKmode.
31662 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31664 * doc/makefile.texi (profiledbootstrap): Refer to the
31665 installation instructions only in textual form.
31667 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31670 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31672 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31674 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31675 (Specific): Update mingw-w64 reference.
31677 (Specific): Remove broken link to Renesas RX processor.
31679 2017-02-10 Richard Biener <rguenther@suse.de>
31681 * toplev.c (process_options): Do not mention obsolete graphite
31682 options when printing sorry message about missing graphite support.
31683 Mention -floop-nest-optimize.
31685 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31687 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31688 (vtst_p16): Likewise.
31689 (vtstq_p8): Likewise.
31690 (vtstq_p16): Likewise.
31692 (vtstq_p64): Likewise.
31693 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31694 (vset_lane_p64): New.
31695 (vsetq_lane_p64): New.
31697 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31699 PR tree-optimization/79411
31700 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31701 stmt operands are SSA_NAMEs used in abnormal phis.
31702 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31705 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31708 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31711 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31713 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31715 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31717 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31721 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31722 not arbitrary TREE_CONSTANT.
31725 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31726 "omp declare target link" attribute unless is_global_var.
31727 * omp-offload.c (find_link_var_op): Likewise.
31729 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31730 Chung-Lin Tang <cltang@codesourcery.com>
31732 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31734 (gimplify_adjust_omp_clauses): Don't delete TILE.
31735 (gimplify_omp_for): Deal with TILE.
31736 * internal-fn.c (expand_GOACC_TILE): New function.
31737 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31739 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31740 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31742 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31743 avoid DIV for outermost collapse var.
31744 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31745 Remove out of date comments, fix whitespace.
31746 * omp-general.c (omp_extract_for_data): Deal with tiling.
31747 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31748 adjust OLF_DIM_BASE value.
31749 (struct omp_for_data): Add tiling field.
31750 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31751 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31752 for auto loops. Remove default auto determining, moved to
31753 oacc_loop_fixed_partitions.
31754 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31755 stmts, add e_mask field.
31756 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31757 (oacc_thread_numbers): Use oacc_dim_call.
31758 (oacc_xform_tile): New.
31759 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31760 (finish_oacc_loop): Adjust for ifns vector.
31761 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31762 add case for GOACC_TILE fns.
31763 (oacc_loop_xform_loop): Delete.
31764 (oacc_loop_process): Iterate over call list directly, and add
31765 handling for GOACC_TILE fns.
31766 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31768 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31769 vector partitioning to outer loops. Assign 2 partitions to loops
31770 when available. Add TILE handling.
31771 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31772 (execite_oacc_device_lower): Process GOACC_TILE fns,
31773 ignore unknown specs.
31774 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31775 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31776 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31778 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31780 * configure.ac (ACX_BUGURL): Update.
31781 * configure: Regenerate.
31783 2017-02-09 Richard Biener <rguenther@suse.de>
31785 PR tree-optimization/69823
31786 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31787 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31789 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31791 * config/arc/arc-c.def: Add __NPS400__ definition.
31792 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31793 (TARGET_NPS400): Define.
31795 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31797 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31799 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31800 pointer, arch_info.
31801 (arc_cpu_types): Fill the arch_info field with a pointer into the
31802 arc_arch_types table.
31803 (arc_selected_cpu): Declare.
31804 * config/arc/arc.c (arc_selected_cpu): Make global.
31805 (arc_selected_arch): Delete.
31806 (arc_base_cpu): Delete.
31807 (arc_override_options): Remove references to deleted variables,
31808 update access to arch information.
31809 (ARC_OPT): Update access to arch information.
31810 (ARC_OPTX): Likewise.
31811 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31812 (TARGET_ARC600): Update access to arch information.
31813 (TARGET_ARC601): Likewise.
31814 (TARGET_ARC700): Likewise.
31815 (TARGET_EM): Likewise.
31816 (TARGET_HS): Likewise.
31817 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31820 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31823 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31824 condition/operands for integer GE/LE/GEU/LEU operations.
31826 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31828 PR translation/79397
31829 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31832 2017-02-08 Martin Jambor <mjambor@suse.cz>
31835 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31836 whether allocation happened.
31837 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31838 nothing was allocated.
31840 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31842 PR tree-optimization/79408
31843 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31844 constant, but SSA_NAME with a known integer range, use the minimum
31845 of that range instead of op1 to determine if modulo can be replaced
31846 with its first operand.
31848 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31850 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31852 2017-02-08 Richard Biener <rguenther@suse.de>
31854 PR tree-optimization/71824
31855 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31856 Check all loops contained in the merged region.
31858 2017-02-07 Andrew Pinski <apinski@cavium.com>
31860 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31862 2017-02-07 Andrew Pinski <apinski@cavium.com>
31864 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31865 (thunderxt88): Likewise.
31866 (thunderxt81): Disable LSE and change v8.1 to v8.
31867 (thunderxt83): Likewise.
31869 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31870 Richard Biener <rguenther@suse.de>
31872 PR middle-end/79399
31873 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31874 type from int to size_t.
31875 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31877 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31879 PR rtl-optimization/79386
31880 * cprop.c (bypass_conditional_jumps): Initialize
31881 bypass_last_basic_block already before splitting bbs after
31882 unconditional traps...
31883 (bypass_conditional_jumps): ... rather than here.
31886 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31887 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31888 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31889 fix -masm=intel patterns.
31891 2017-02-07 Richard Biener <rguenther@suse.de>
31893 PR tree-optimization/79256
31894 PR middle-end/79278
31895 * builtins.c (get_object_alignment_2): Use min_align_of_type
31896 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31897 and ADJUST_FIELD_ALIGN.
31899 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31901 * doc/tm.texi: Regenerate.
31902 * stor-layout.c (layout_decl): Adjust.
31903 (update_alignment_for_field): Likewise.
31904 (place_field): Likewise.
31905 (min_align_of_type): Likewise.
31906 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31907 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31908 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31909 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31910 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31911 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31912 * config/i386/i386.c (x86_field_alignment): Likewise.
31913 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31914 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31915 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31916 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31917 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31918 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31922 2017-01-30 Richard Biener <rguenther@suse.de>
31924 PR tree-optimization/79256
31925 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31926 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31929 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31931 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31932 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31933 builtins to SImode and emit a zero-extend, if necessary.
31935 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31937 * docs/invoke.texi (RISC-V Options): Alphabetize.
31939 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31941 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31944 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31946 * config/riscv/riscv.c: New file.
31947 * common/config/riscv/riscv-common.c: Likewise.
31948 * config.gcc: Likewise.
31949 * config/riscv/constraints.md: Likewise.
31950 * config/riscv/elf.h: Likewise.
31951 * config/riscv/generic.md: Likewise.
31952 * config/riscv/linux.h: Likewise.
31953 * config/riscv/multilib-generator: Likewise.
31954 * config/riscv/peephole.md: Likewise.
31955 * config/riscv/pic.md: Likewise.
31956 * config/riscv/predicates.md: Likewise.
31957 * config/riscv/riscv-builtins.c: Likewise.
31958 * config/riscv/riscv-c.c: Likewise.
31959 * config/riscv/riscv-ftypes.def: Likewise.
31960 * config/riscv/riscv-modes.def: Likewise.
31961 * config/riscv/riscv-opts.h: Likewise.
31962 * config/riscv/riscv-protos.h: Likewise.
31963 * config/riscv/riscv.h: Likewise.
31964 * config/riscv/riscv.md: Likewise.
31965 * config/riscv/riscv.opt: Likewise.
31966 * config/riscv/sync.md: Likewise.
31967 * config/riscv/t-elf-multilib: Likewise.
31968 * config/riscv/t-linux: Likewise.
31969 * config/riscv/t-linux-multilib: Likewise.
31970 * config/riscv/t-riscv: Likewise.
31971 * configure.ac: Likewise.
31972 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31973 Waterman as RISC-V maintainers.
31974 * doc/install.texi: Add RISC-V entries.
31975 * doc/invoke.texi: Add RISC-V options section.
31976 * doc/md.texi: Add RISC-V constraints section.
31977 * configure: Regenerated.
31979 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31982 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31983 false values to be constant vectors with all 0 or all 1 bits set.
31984 (vcondu<mode><mode>): Likewise.
31985 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31987 (fpmask_comparison_operator): Update comment.
31988 (vecint_comparison_operator): New predicate.
31989 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31990 vector conditionals when the true and false values are constant
31991 vectors with all 0 bits or all 1 bits set.
31993 2017-02-06 Martin Sebor <msebor@redhat.com>
31995 PR tree-optimization/79376
31996 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31998 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32000 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32001 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32002 to simplify split condition.
32004 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32006 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32007 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32010 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32012 PR rtl-optimization/68664
32013 * target.def (can_speculate_insn): New hook.
32014 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32015 * doc/tm.texi: Regenerate.
32016 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32017 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32018 (rs6000_sched_can_speculate_insn): New function.
32020 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32022 PR tree-optimization/79284
32023 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32024 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32025 vectorizable_mask_load_store, vectorizable_operation,
32026 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32027 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32028 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32029 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32030 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32031 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32032 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32033 is_gimple_assign (stmt). Replace another such test with
32034 is_gimple_assign (stmt).
32036 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32039 * config/avr/avr.c (rtl-iter.h): Include it.
32040 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32041 (avr_legitimate_combined_insn): ...and implementation.
32043 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32045 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32046 * config/s390/s390.c (s390_const_operand_ok)
32047 (s390_canonicalize_comparison, s390_extract_part)
32048 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32049 (s390_contiguous_bitmask_p, s390_rtx_costs)
32050 (legitimize_pic_address): Likewise.
32051 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32052 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32053 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32054 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32055 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32057 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32059 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32060 REGNO($0) == REGNO($1).
32062 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32064 * config/s390/linux.h(SIZE_TYPE): Add comment.
32066 2017-02-06 Julian Brown <julian@codesourcery.com>
32067 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32068 Virendra Pathak <virendra.pathak@broadcom.com>
32070 * config/aarch64/aarch64-cores.def: Change the scheduler
32072 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32073 * config/aarch64/thunderx2t99.md: New file.
32075 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32077 * doc/standards.texi (Go Language): Update link to language
32080 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32082 * tree-eh.c (lower_resx): Sanitize profile.
32083 (cleanup_empty_eh_move_lp): Likewise.
32085 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32088 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32090 * cfgloopmanip.h (loop_version): Update prototype.
32091 * modulo-sched.c (sms_schedule): Update call of loop_version.
32092 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32093 * tree-parloops.c (gen_parallel_loop): Likewise.
32094 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32095 * tree-ssa-loop-split.c (split_loop): Likewise.
32096 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32097 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32099 2017-02-05 Martin Liska <mliska@suse.cz>
32102 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32104 (print_operand_address): Initialize a struct to zero.
32106 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32108 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32109 garbage collector only in textual form.
32111 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32113 * doc/extend.texi (x86 specific memory model extensions for
32114 transactional memory): Simplify a phrase.
32116 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32119 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32120 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32121 (atomic_storedi_1): Likewise.
32123 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32125 PR tree-optimization/79338
32126 * tree-parloops.c (gather_scalar_reductions): Don't call
32127 vect_analyze_loop_form for loop->inner before destroying loop's
32130 2017-02-03 Martin Sebor <msebor@redhat.com>
32132 PR tree-optimization/79327
32133 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32134 when precision has resulted in leading zeros.
32135 (format_integer): Adjust the likely counter to assume an unknown
32136 argument that may be zero is non-zero.
32138 2017-02-03 Jason Merrill <jason@redhat.com>
32141 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32142 avoid copying non-taken branch.
32144 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32146 PR tree-optimization/79340
32147 * tree-vect-loop.c (vectorizable_reduction): Release
32148 vec_defs elements after safe_splicing them into other vectors.
32151 PR tree-optimization/79327
32152 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32153 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32155 (format_integer): Use wide_int_to_tree instead of build_int_cst
32156 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32157 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32158 of shortest and longest sequence.
32160 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32162 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32163 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32165 2017-02-03 Walter Lee <walt@tilera.com>
32168 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32169 after initial stackframe link reg save.
32170 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32172 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32175 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32176 wu for stxssp alternative.
32178 2017-02-03 Martin Sebor <msebor@redhat.com>
32180 PR tree-optimization/79352
32181 * gimple-fold.c (get_range_strlen): Add argument.
32182 (get_range_strlen): Change return type to bool.
32183 (get_maxval_strlen): Pass in a dummy argument.
32184 * gimple-fold.h (get_range_strlen): Change return type to bool.
32185 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32186 * tree.h (array_at_struct_end_p): Add argument.
32187 * tree.c (array_at_struct_end_p): Handle it.
32189 2017-02-03 Martin Liska <mliska@suse.cz>
32192 * multiple_target.c (create_dispatcher_calls): Redirect edge
32193 from a caller of a dispatcher.
32194 (expand_target_clones): Make the clones local.
32195 (ipa_target_clone): Do both target clones and resolvers.
32196 (ipa_dispatcher_calls): Remove the pass.
32197 (pass_dispatcher_calls::gate): Likewise.
32198 (make_pass_dispatcher_calls): Likewise.
32199 * passes.def (pass_target_clone): Put as very first IPA early
32202 2017-02-03 Martin Liska <mliska@suse.cz>
32204 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32205 in case of a function with ifunc attribute.
32207 2017-02-03 Martin Liska <mliska@suse.cz>
32209 * cgraph.c (cgraph_node::dump): Dump function version info.
32210 * symtab.c (symtab_node::dump_base): Add missing new line.
32212 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32214 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32215 (ifcombine_ifandif): Use it.
32217 2017-02-03 Martin Liska <mliska@suse.cz>
32219 * doc/invoke.texi: Document default value for
32220 use-after-scope-direct-emission-threshold.
32222 2017-02-03 Martin Liska <mliska@suse.cz>
32224 PR tree-optimization/79339
32225 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32226 (format_floating): Likewise.
32228 2017-02-03 Martin Liska <mliska@suse.cz>
32231 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32233 (ipa_node_params_t::remove): Likewise.
32234 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32235 initialization from removed ipa_node_params_t::insert.
32236 (ipa_node_params::~ipa_node_params): Move from removed
32237 ipa_node_params_t::release.
32238 * symbol-summary.h (symbol_summary::m_released): New member.
32239 Do not release a summary twice. Do not allow to call finalizer
32240 for types of a summary that live in GGC memory.
32242 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32244 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32247 2017-02-02 Martin Sebor <msebor@redhat.com>
32249 PR middle-end/79275
32250 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32251 (format_string): Tighten up the range of output for non-constant
32252 strings and correct the expected range for wide non-constant strings.
32254 2017-02-02 Martin Sebor <msebor@redhat.com>
32256 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32258 PR middle-end/32003
32259 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32261 (-fdump-tree-@var): Add to index and document how to come up
32262 with pass-specific option and dump file names.
32263 (-fdump-passes): Clarify where to look for output.
32265 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32267 PR middle-end/77445
32268 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32269 statistics of the analyzed path; allow threading for speed when
32270 any of BBs along the path are optimized for speed.
32272 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32274 PR middle-end/78468
32275 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32276 settings of the virtual registers.
32279 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32281 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32282 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32285 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32287 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32288 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32290 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32292 * config/s390/s390.md: Add missing comments with the expanded
32294 * config/s390/vector.md: Likewise.
32295 * config/s390/vx-builtins.md: Likewise.
32297 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32300 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32301 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32302 conditions on a single line.
32304 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32306 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32307 __S390_VX__ to __VX__.
32309 2017-02-01 Andrew Pinski <apinski@cavium.com>
32311 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32312 stmt_info to record_stmt_cost.
32313 (vect_get_known_peeling_cost): Pass stmt_info if known to
32315 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32316 cpu_vector_cost field into
32317 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32318 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32319 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32320 splitting of scalar_stmt_cost and vec_stmt_cost.
32321 (thunderx_vector_cost): Likewise.
32322 (cortexa57_vector_cost): LIkewise.
32323 (exynosm1_vector_cost): Likewise.
32324 (xgene1_vector_cost): Likewise.
32325 (thunderx2t99_vector_cost): Improve after the splitting of the two
32327 (aarch64_builtin_vectorization_cost): Update for the splitting of
32328 scalar_stmt_cost and vec_stmt_cost.
32330 2017-02-01 Torvald Riegel <triegel@redhat.com>
32331 Richard Henderson <rth@redhat.com>
32333 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32334 conditional on existance of a fast atomic load.
32335 * optabs-query.c (can_atomic_load_p): New function.
32336 * optabs-query.h (can_atomic_load_p): Declare it.
32337 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32338 no fast atomic load is available for the particular size of access.
32339 (expand_atomic_compare_and_swap): Likewise.
32340 (expand_atomic_load): Likewise.
32341 (expand_atomic_store): Likewise.
32342 (expand_atomic_fetch_op): Likewise.
32343 * testsuite/lib/target-supports.exp
32344 (check_effective_target_sync_int_128): Remove x86 because it provides
32345 no fast atomic load.
32346 (check_effective_target_sync_int_128_runtime): Likewise.
32348 2017-02-01 Richard Biener <rguenther@suse.de>
32350 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32351 (graphite_transform_loops): Provide opt-info for optimized nests.
32352 * tree-parloop.c (parallelize_loops): Provide opt-info for
32353 parallelized loops.
32355 2017-02-01 Richard Biener <rguenther@suse.de>
32357 PR middle-end/79315
32358 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32359 was not set before.
32361 2017-02-01 Richard Biener <rguenther@suse.de>
32363 PR tree-optimization/71824
32364 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32365 Verify the loops are valid in the merged SESE region.
32366 (scop_detection::can_represent_loop_1): Check analyzing the
32367 evolution of the number of iterations in the region succeeds.
32369 2017-01-31 Ian Lance Taylor <iant@golang.org>
32371 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32372 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32374 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32376 PR preprocessor/79210
32377 * input.c (get_substring_ranges_for_loc): Replace line_width
32378 assertion with error-handling.
32380 2017-01-31 Richard Biener <rguenther@suse.de>
32382 PR tree-optimization/77318
32383 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32384 (create_pw_aff_from_tree): Take loop parameter.
32385 (add_condition_to_pbb): Pass loop of the condition to
32386 create_pw_aff_from_tree.
32388 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32390 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32391 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32393 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32397 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32399 (convert_int_to_float128): Likewise.
32400 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32401 (convert_int_to_float128): Likewise.
32402 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32403 (UNSPEC_IEEE128_CONVERT): Likewise.
32404 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32405 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32406 Use local variables for IBM extended format.
32407 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32408 (fix_trunc<mode>si2_fprs): Likewise.
32409 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32410 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32411 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32412 to know that we can now have integers of all sizes in vector
32414 (fix<uns>_<mode>di2_hw): Likewise.
32415 (float<uns>_<mode>si2_hw): Likewise.
32416 (fix_<mode>si2_hw): Likewise.
32417 (fixuns_<mode>si2_hw): Likewise.
32418 (float<uns>_<mode>di2_hw): Likewise.
32419 (float_<mode>di2_hw): Likewise.
32420 (float_<mode>si2_hw): Likewise.
32421 (floatuns_<mode>di2_hw): Likewise.
32422 (floatuns_<mode>si2_hw): Likewise.
32423 (xscvqp<su>wz_<mode>): Delete, no longer used.
32424 (xscvqp<su>dz_<mode>): Likewise.
32425 (xscv<su>dqp_<mode>): Likewise.
32426 (ieee128_mfvsrd_64bit): Likewise.
32427 (ieee128_mfvsrd_32bit): Likewise.
32428 (ieee128_mfvsrwz): Likewise.
32429 (ieee128_mtvsrw): Likewise.
32430 (ieee128_mtvsrd_64bit): Likewise.
32431 (ieee128_mtvsrd_32bit): Likewise.
32433 2017-01-31 Martin Liska <mliska@suse.cz>
32436 * ipa-prop.c (ipa_free_all_node_params): Call release method
32437 instead of ~sumbol_summary to not to trigger double times
32440 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32442 PR tree-optimization/71691
32443 * bitmap.h (class auto_bitmap): New.
32444 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32445 is_maybe_undefined instead of ssa_undefined_value_p.
32447 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32449 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32450 __S390_ARCH_LEVEL__ to __ARCH__.
32452 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32454 PR tree-optimization/79267
32455 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32456 if should_remove_lhs_p is true.
32458 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32461 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32462 (add_alignment_attribute): New.
32463 (base_type_die): Add alignment attribute.
32464 (subrange_type_die): Likewise.
32465 (modified_type_die): Likewise.
32466 (gen_array_type_die): Likewise.
32467 (gen_descr_array_type_die: Likewise.
32468 (gen_enumeration_type_die): Likewise.
32469 (gen_subprogram_die): Likewise.
32470 (gen_variable_die): Likewise.
32471 (gen_field_die): Likewise.
32472 (gen_ptr_to_mbr_type_die): Likewise.
32473 (gen_struct_or_union_type_die): Likewise.
32474 (gen_subroutine_type_die): Likewise.
32475 (gen_typedef_die): Likewise.
32476 (base_type_cmp): Compare alignment attribute.
32478 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32481 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32482 (setb_unsigned) New pattern for setb with CCUNS.
32483 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32484 subfc./subfe sequence to avoid overflow problems. Generate a
32485 shorter sequence with cmpld/setb for power9.
32486 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32487 for generating subfc. instruction.
32488 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32489 now uses this instruction.
32491 2017-01-30 Ian Lance Taylor <iant@google.com>
32494 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32495 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32497 2017-01-30 Martin Sebor <msebor@redhat.com>
32499 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32500 Move constant to the right of a relational operator.
32501 (get_mpfr_format_length, format_character, format_string): Ditto.
32502 (should_warn_p, maybe_warn): Same.
32504 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32506 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32509 * asan.c (get_translation_unit_decl): Remove function.
32510 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32512 2017-01-30 Martin Liska <mliska@suse.cz>
32514 PR gcov-profile/79259
32515 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32516 -fprofile-generate.
32518 2017-01-30 Martin Liska <mliska@suse.cz>
32521 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32522 Initialize variables with NULL value.
32524 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32527 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32529 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32531 2017-01-30 Richard Biener <rguenther@suse.de>
32533 PR tree-optimization/79276
32534 * tree-vrp.c (process_assert_insertions): Properly adjust common
32535 when removing a duplicate.
32537 2017-01-30 Richard Biener <rguenther@suse.de>
32539 PR tree-optimization/79256
32540 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32541 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32543 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32545 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32548 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32549 ("*r<noxa>sbg_<mode>_sll_bitmask")
32550 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32551 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32552 Use contiguous_bitmask_nowrap_operand.
32554 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32557 * config/rs6000/altivec.h (vec_xl): Revise #define.
32558 (vec_xst): Likewise.
32560 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32562 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32564 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32566 PR rtl-optimization/79194
32567 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32568 traps before call to bypass_conditional_jumps.
32570 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32572 PR tree-optimization/71374
32573 * lra-constraints.c (check_conflict_input_operands): New.
32574 (match_reload): Use it.
32576 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32579 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32580 account to calculate conflict_set.
32582 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32584 PR rtl-optimization/78559
32585 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32586 other_insn in combine.
32588 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32590 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32591 uint16_type_node for BT_UINT16.
32593 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32595 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32596 "RTL Tests" to menu.
32597 (GIMPLE Tests): New node.
32598 (RTL Tests): New node.
32600 2017-01-27 Richard Biener <rguenther@suse.de>
32602 PR tree-optimization/79245
32603 * tree-loop-distribution.c (distribute_loop): Apply cost
32604 modeling also to detected patterns.
32606 2017-01-27 Richard Biener <rguenther@suse.de>
32608 PR tree-optimization/71433
32609 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32610 (compare_assert_loc): New function.
32611 (process_assert_insertions): Sort and optimize assert locations
32612 to remove duplicates and push down identical assertions on
32613 edges to their destination block.
32615 2017-01-27 Richard Biener <rguenther@suse.de>
32617 PR tree-optimization/79244
32618 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32619 out SSA names even if abnormal.
32621 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32623 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32624 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32625 instead of MPFR_RNDN.
32627 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32630 * arm.c (arm_option_override): Don't call build_target_option_node
32631 until after doing all option overrides.
32632 (arm_valid_target_attribute_tree): Likewise.
32634 2017-01-27 Martin Liska <mliska@suse.cz>
32636 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32637 for {cd}tors and C++ {cd}tors.
32639 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32641 * config/s390/s390.md ("*setmem_long_and")
32642 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32644 2017-01-26 Martin Sebor <msebor@redhat.com>
32646 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32649 2017-01-26 Martin Sebor <msebor@redhat.com>
32651 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32652 HAVE_DFmode before using XFmode or DFmode.
32653 (parse_directive): Avoid using the z length modifier to avoid
32654 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32656 PR middle-end/78703
32657 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32658 to accept adjustment as an array.
32659 (get_int_range): New function.
32660 (struct directive): Make width and prec arrays.
32661 (directive::set_width, directive::set_precision): Call get_int_range.
32662 (format_integer, format_floating): Handle width and precision ranges.
32663 (format_string, parse_directive): Same.
32665 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32668 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32669 comdat_type_p set, just clone them, but keep the children in the
32673 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32674 which have direct callers with -fvar-tracking-assignments enabled
32676 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32677 inside of type units.
32679 2017-01-26 Martin Sebor <msebor@redhat.com>
32681 PR middle-end/78703
32682 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32684 (struct format_result): Replace number_chars, number_chars_min,
32685 and number_chars_max with a single member of struct result_range.
32687 (format_result::operator+=): Adjust.
32688 (struct fmtresult): Remove bounded. Handle likely and unlikely
32690 (fmtresult::adjust_for_width_or_precision): New function.
32691 (fmtresult:type_max_digits): New function.
32692 (bytes_remaining): Handle likely and unlikely counters.
32693 (min_bytes_remaining): Remove.
32694 (format_percent): Simplify.
32695 (format_integer, format_floating): Set likely and unlikely counters.
32696 (get_string_length, format_character, format_string): Same.
32697 (format_plain, should_warn_p): New function.
32698 (maybe_warn): Call should_warn_p. Update diagnostic messages
32699 and handle those for all directives, including plain strings.
32700 (format_directive): Handle likely and unlikely counters.
32701 Remove unnecessary quoting from diagnostics. Add an informational
32703 (add_bytes): Remove.
32704 (pass_sprintf_length::compute_format_length): Simplify.
32705 (try_substitute_return_value): Handle likely and unlikely counters.
32707 2017-01-26 Carl Love <cel@us.ibm.com>
32709 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32710 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32712 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32715 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32716 endianess for subregs into account.
32717 * lra-constraints.c (lra_constraints): Do risky transformations
32718 always on the first iteration.
32719 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32720 last_call_used_reg_set.
32721 (process_bb_lives): Define and use last_call_used_reg_set.
32722 * lra.c (lra): Always continue after lra_constraints on the first
32725 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32727 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32729 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32731 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32733 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32734 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32735 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32736 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32737 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32738 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32739 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32740 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32741 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32743 2017-01-26 Marek Polacek <polacek@redhat.com>
32746 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32747 for the third operand.
32749 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32751 PR middle-end/79236
32752 * omp-low.c (struct omp_context): Add simt_stmt field.
32753 (scan_omp_for): Return omp_context *.
32754 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32755 context to the _simt_ SIMD stmt.
32756 (lower_omp_for): For combined SIMD with sibling _simt_
32757 SIMD, make sure to use the same decls in _looptemp_
32758 clauses as in the sibling.
32760 2017-01-26 David Sherwood <david.sherwood@arm.com>
32762 PR middle-end/79212
32763 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32766 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32769 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32770 emit fld b; fld a; if possible.
32772 * brig-builtins.def: Update copyright years.
32773 * config/arm/arm_acle_builtins.def: Update copyright years.
32775 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32778 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32779 constraint instead of o for the stxsd instruction.
32781 2017-01-25 Carl Love <cel@us.ibm.com>
32783 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32784 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32786 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32788 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32790 2017-01-25 Richard Biener <rguenther@suse.de>
32792 PR tree-optimization/69264
32793 * target.def (vector_alignment_reachable): Improve documentation.
32794 * doc/tm.texi: Regenerate.
32795 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32797 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32798 earlier changes with respect to TYPE_USER_ALIGN.
32799 (vector_alignment_reachable_p): Likewise. Improve dumping.
32801 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32804 * config/arm/arm.md (xordi3): Force constant operand into a register
32807 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32809 * doc/invoke.texi (-fstore-merging): Correct default optimization
32810 levels at which it is enabled.
32811 (-O): Move -fstore-merging from list to...
32814 2017-01-25 Richard Biener <rguenther@suse.de>
32817 * omp-expand.c: Include debug.h.
32818 (expand_omp_taskreg): Make sure to generate early debug before
32819 outlining anything from a function.
32820 (expand_omp_target): Likewise.
32821 (grid_expand_target_grid_body): Likewise.
32823 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32826 * asan.c (get_translation_unit_decl): New function.
32827 (asan_add_global): Extract modules file name from globals
32828 TRANSLATION_UNIT_DECL name.
32830 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32833 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32834 for long calls with APCS frame and VFP.
32836 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32838 * cfg.c (original_copy_tables_initialized_p): New function.
32839 * cfg.h (original_copy_tables_initialized_p): New decl.
32840 * cfgrtl.c (relink_block_chain): Guard the call to
32841 free_original_copy_tables with a call to
32842 original_copy_tables_initialized_p.
32843 * cgraph.h (symtab_node::native_rtl_p): New decl.
32844 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32845 (symtab_node::needed_p): Don't assert for early assembly output
32846 for __RTL functions.
32847 (cgraph_node::finalize_function): Set "force_output" for __RTL
32849 (cgraph_node::analyze): Bail out early for __RTL functions.
32850 (analyze_functions): Update assertion to support __RTL functions.
32851 (cgraph_node::expand): Bail out early for __RTL functions.
32852 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32854 * function.h (struct function): Update comment for field
32856 * gimple-expr.c: Include "tree-pass.h".
32857 (gimple_has_body_p): Return false for __RTL functions.
32858 * Makefile.in (OBJS): Add run-rtl-passes.o.
32859 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32861 (gcc::pass_manager::get_clean_slate): New accessor.
32862 * passes.c: Include "insn-addr.h".
32863 (should_skip_pass_p): Add logging. Update logic for running
32864 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32865 property-provider override so it is only done for gimple passes.
32867 (skip_pass): New function.
32868 (execute_one_pass): Call skip_pass when skipping passes.
32869 * read-md.c (md_reader::read_char): Support filtering
32870 the input to a subset of line numbers.
32871 (md_reader::md_reader): Initialize fields
32872 m_first_line and m_last_line.
32873 (md_reader::read_file_fragment): New function.
32874 * read-md.h (md_reader::read_file_fragment): New decl.
32875 (md_reader::m_first_line): New field.
32876 (md_reader::m_last_line): New field.
32877 * read-rtl-function.c (function_reader::create_function): Only
32878 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32879 curr_properties. Set DECL_INITIAL to a dummy block.
32880 (read_rtl_function_body_from_file_range): New function.
32881 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32883 * run-rtl-passes.c: New file.
32884 * run-rtl-passes.h: New file.
32886 2017-01-24 Jeff Law <law@redhat.com>
32888 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32891 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32893 PR tree-optimization/79159
32894 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32895 (record_nonwrapping_iv): Improve boundary using above function if no
32896 value range information.
32898 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32899 Martin Jambor <mjambor@suse.cz>
32901 * brig-builtins.def: New file.
32902 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32903 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32904 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32905 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32906 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32907 * builtin-types.def (BT_INT8): New.
32908 (BT_INT16): Likewise.
32909 (BT_UINT8): Likewise.
32910 (BT_UINT16): Likewise.
32911 (BT_FN_ULONG): Likewise.
32912 (BT_FN_UINT_INT): Likewise.
32913 (BT_FN_UINT_ULONG): Likewise.
32914 (BT_FN_UINT_LONG): Likewise.
32915 (BT_FN_UINT_PTR): Likewise.
32916 (BT_FN_ULONG_PTR): Likewise.
32917 (BT_FN_INT8_FLOAT): Likewise.
32918 (BT_FN_INT16_FLOAT): Likewise.
32919 (BT_FN_UINT32_FLOAT): Likewise.
32920 (BT_FN_UINT16_FLOAT): Likewise.
32921 (BT_FN_UINT8_FLOAT): Likewise.
32922 (BT_FN_UINT64_FLOAT): Likewise.
32923 (BT_FN_UINT16_UINT32): Likewise.
32924 (BT_FN_UINT32_UINT16): Likewise.
32925 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32926 (BT_FN_INT_PTR_INT): Likewise.
32927 (BT_FN_UINT_PTR_UINT): Likewise.
32928 (BT_FN_LONG_PTR_LONG): Likewise.
32929 (BT_FN_ULONG_PTR_ULONG): Likewise.
32930 (BT_FN_VOID_UINT64_UINT64): Likewise.
32931 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32932 (BT_FN_INT8_INT8_INT8): Likewise.
32933 (BT_FN_INT16_INT16_INT16): Likewise.
32934 (BT_FN_INT_INT_INT): Likewise.
32935 (BT_FN_UINT_FLOAT_UINT): Likewise.
32936 (BT_FN_FLOAT_UINT_UINT): Likewise.
32937 (BT_FN_ULONG_UINT_UINT): Likewise.
32938 (BT_FN_ULONG_UINT_PTR): Likewise.
32939 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32940 (BT_FN_UINT_UINT_UINT): Likewise.
32941 (BT_FN_VOID_UINT_PTR): Likewise.
32942 (BT_FN_UINT_UINT_PTR: Likewise.
32943 (BT_FN_UINT32_UINT64_PTR): Likewise.
32944 (BT_FN_INT_INT_UINT_UINT): Likewise.
32945 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32946 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32947 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32948 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32949 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32950 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32951 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32952 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32953 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32954 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32955 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32956 * doc/frontends.texi: List BRIG FE.
32957 * doc/install.texi (Testing): Add BRIG tesring requirements.
32958 * doc/invoke.texi (Overall Options): Mention BRIG.
32959 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32961 2017-01-24 Richard Biener <rguenther@suse.de>
32963 PR translation/79208
32964 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32966 2017-01-24 Martin Jambor <mjambor@suse.cz>
32969 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32970 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32971 and known_contexts.
32973 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32975 PR middle-end/79123
32976 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32977 casts from signed to unsigned really don't have a range.
32979 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32981 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32982 GMP_RNDx for compatiblity.
32984 2017-01-24 Martin Liska <mliska@suse.cz>
32987 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32988 that would prevent us to call alloca with -1 as argument.
32990 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32992 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32993 Avoid -Wformat-security warning.
32995 2017-01-23 Andrew Pinski <apinski@cavium.com>
32997 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33000 2017-01-23 Martin Sebor <msebor@redhat.com>
33002 PR middle-end/78703
33003 * gimple-ssa-sprintf.c (warn_level): New global.
33004 (format_integer): Use it here and throughout the rest of the file.
33005 Use the same switch to compute sign as base.
33006 (maybe_warn): New function.
33007 (format_directive): Factor out warnings into maybe_warn.
33008 Add debugging output. Use warn_level.
33009 (add_bytes): Use warn_level.
33010 (pass_sprintf_length::compute_format_length): Add debugging output.
33011 (try_substitute_return_value): Same.
33012 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33014 PR middle-end/78703
33015 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33016 (struct fmtresult, format_integer, format_floating): Adjust.
33017 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33018 (get_string_length, format_string,format_directive): Same.
33019 (pass_sprintf_length::compute_format_length): Same.
33020 (try_substitute_return_value): Simplify slightly.
33022 PR middle-end/78703
33023 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33024 (fmtresult::operator+=): Outlined.
33025 (struct fmtresult): Add ctors.
33026 (struct conversion_spec): Rename...
33027 (struct directive): ...to this. Add and remove data members.
33028 (directive::set_width, directive::set_precision): New functions.
33029 (format_percent): Use fmtresult ctor.
33030 (get_width_and_precision): Remove.
33031 (format_integer): Make naming changes. Avoid computing width and
33033 (format_floating): Same. Adjust indentation.
33034 (format_character, format_none): New functions.
33035 (format_string): Moved character handling to format_character.
33036 (format_directive): Remove arguments, change return type.
33037 (parse_directive): New function.
33038 (pass_sprintf_length::compute_format_length): Move directive
33039 parsing to parse_directive.
33041 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33043 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33044 (assign_assembler_name_if_needed): ... this.
33045 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33046 (assign_assembler_name_if_needed): ... this.
33047 (free_lang_data_in_cgraph): Adjust callers.
33048 * cgraphunit.c (cgraph_node::analyze): Likewise.
33049 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33052 2017-01-23 Richard Biener <rguenther@suse.de>
33054 PR tree-optimization/79088
33055 PR tree-optimization/79188
33056 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33057 resetting loop bounds after last path deletion. Reset loop
33058 bounds of the target loop, make code match the comments.
33059 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33060 Make sure loops need no fixups.
33062 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33064 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33065 exponent support with double type for first argument.
33066 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33067 type returned by __builtin_vec_extract_sig,
33068 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33069 functions from "vector int" to "vector unsigned int" or from
33070 "vector long long int" to "vector unsigned long long int".
33071 Changed type returned by __builtin_vec_extract_exp,
33072 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33073 functions from "vector int" to "vector unsigned int" or from
33074 "vector long long int" to "vector unsigned long long int".
33075 Changed return type of __builtin_vec_test_data_class,
33076 __builtin_vec_test_data_class_sp, and
33077 __builtin_vec_test_data_class_dp from "vector int" to
33078 "vector bool int" or from "vector long long int" to "vector bool
33079 long long int" and changed second argument type from "unsigned
33080 int" to "int". Added new overloaded function forms "vector float
33081 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33082 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33083 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33084 double, vector unsigned long long int)" and "vector double
33085 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33086 long int)". Changed return type of
33087 __builtin_scalar_test_data_class and
33088 __builtin_scalar_test_data_class_sp and
33089 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33090 int" and changed second argument from "unsigned int" to "int".
33091 Changed type returned by __builtin_scalar_test_neg,
33092 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33093 from "int" to "bool int". Added new overloaded function form
33094 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33095 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33096 exponent double-precision with floating point first argument.
33097 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33098 documentation of scalar_test_data_class, scalar_test_neg,
33099 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33100 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33101 vec_test_data_class built-in functions to reflect refinements in
33102 their type signatures.
33104 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33106 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33108 (aarch64_elf_asm_destructor): Likewise.
33110 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33112 PR rtl-optimization/78634
33113 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33114 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33115 * ifcvt.c (noce_try_cmove): Add missing cost check.
33117 PR rtl-optimization/71724
33118 * combine.c (if_then_else_cond): Look for situations where it is
33119 beneficial to undo the work of one of the recursive calls.
33121 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33123 PR tree-optimization/70754
33124 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33125 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33126 combined stmt before it if not NULL.
33127 (combine_chains): Process refs reversely and compute dominance point
33130 2017-01-23 Martin Liska <mliska@suse.cz>
33132 PR tree-optimization/79196
33133 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33134 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33136 (strlen_optimize_stmt): Call the renamed function.
33138 2017-01-23 Michael Matz <matz@suse.de>
33140 PR tree-optimization/78384
33141 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33143 2017-01-23 Richard Biener <rguenther@suse.de>
33145 PR tree-optimization/79186
33146 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33147 both incoming edges before moving an assert.
33149 2017-01-23 Martin Jambor <mjambor@suse.cz>
33151 * ipa-prop.c (load_from_param_1): Removed.
33152 (load_from_unmodified_param): Bits from load_from_param_1 put back
33154 (load_from_param): Removed.
33155 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33156 with stmt. Reverted back to use of load_from_unmodified_param.
33158 2017-01-23 Martin Jambor <mjambor@suse.cz>
33161 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33162 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33163 field a pointer to garbage collected vector, mark lattices and
33164 ipcp_orig_node with GTY((skip)).
33165 (ipa_get_param_count): Adjust to descriptors being a pointer.
33166 (ipa_get_param): Likewise.
33167 (ipa_get_type): Likewise.
33168 (ipa_get_param_move_cost): Likewise.
33169 (ipa_set_param_used): Likewise.
33170 (ipa_get_controlled_uses): Likewise.
33171 (ipa_set_controlled_uses): Likewise.
33172 (ipa_is_param_used): Likewise.
33173 (ipa_node_params_t): Move into garbage collector. New methods insert
33175 (ipa_node_params_sum): Annotate wth GTY(()).
33176 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33178 (ipa_load_from_parm_agg): Adjust declaration.
33179 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33180 * ipa-profile.c (ipa_profile): Likewise.
33181 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33182 (ipa_populate_param_decls): Make descriptors parameter garbage
33184 (ipa_dump_param): Adjust to descriptors being a pointer.
33185 (ipa_alloc_node_params): Likewise.
33186 (ipa_initialize_node_params): Likewise.
33187 (load_from_param_1): Make descriptors parameter garbage collected.
33188 (load_from_unmodified_param): Likewise.
33189 (load_from_param): Likewise.
33190 (ipa_load_from_parm_agg): Likewise.
33191 (ipa_node_params::~ipa_node_params): Removed.
33192 (ipa_free_all_node_params): Remove call to delete operator.
33193 (ipa_node_params_t::insert): New.
33194 (ipa_node_params_t::remove): Likewise.
33195 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33196 copy known_csts and known_contexts vectors.
33197 (ipa_read_node_info): Adjust to descriptors being a pointer.
33198 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33200 (ipcp_transform_function): Make descriptors variable garbage
33203 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33205 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33206 * config/i386/avx512dqintrin.h: Ditto.
33207 * config/i386/avx512fintrin.h: Ditto.
33208 * config/i386/i386.c: Handle new builtins.
33209 * config/i386/i386-builtin.def: Add new builtins.
33210 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33211 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33213 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33214 Martin Liska <mliska@suse.cz>
33216 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33217 * asan.c (asan_expand_poison_ifn): Support stores and use
33218 appropriate ASAN report function.
33219 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33220 * internal-fn.def (ASAN_POISON_USE): Declare.
33221 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33222 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33223 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33224 ASAN_POISON calls w/o LHS.
33225 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33226 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33227 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33228 * gimplify.c (asan_poison_variables): Add attribute
33229 use_after_scope_memory to variables that really needs to live
33231 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33232 having the attribute.
33234 2017-01-23 Martin Liska <mliska@suse.cz>
33236 * asan.c (create_asan_shadow_var): New function.
33237 (asan_expand_poison_ifn): Likewise.
33238 * asan.h (asan_expand_poison_ifn): New declaration.
33239 * internal-fn.c (expand_ASAN_POISON): Likewise.
33240 * internal-fn.def (ASAN_POISON): New builtin.
33241 * sanopt.c (pass_sanopt::execute): Expand
33242 asan_expand_poison_ifn.
33243 * tree-inline.c (copy_decl_for_dup_finish): Make function
33245 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33246 * tree-ssa.c (is_asan_mark_p): New function.
33247 (execute_update_addresses_taken): Rewrite local variables
33248 (identified just by use-after-scope as addressable) into SSA.
33250 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33252 * doc/install.texi (Specific): opensource.apple.com uses https
33253 now. Remove trailing slash.
33255 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33257 * README.Portability: Remove note on an Irix compatibility issue.
33259 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33261 * gcov.c (INCLUDE_ALGORITHM): Define.
33262 (INCLUDE_VECTOR): Define.
33263 No longer include <vector> and <algorithm> directly.
33265 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33267 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33269 * doc/invoke.texi (Code Gen Options): Ditto.
33271 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33274 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33276 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33278 rtl-optimization/79125
33279 * cprop.c (local_cprop_pass): Handle cases where we make an
33280 unconditional trap.
33282 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33286 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33287 read from, for big endian.
33289 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33291 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33292 register pauth builtins for LP64 only.
33294 2017-01-20 Marek Polacek <polacek@redhat.com>
33297 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33300 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33302 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33304 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33305 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33306 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33308 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33311 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33312 in big-endian mode when they are not a single duplicated value.
33314 2017-01-20 Richard Biener <rguenther@suse.de>
33316 * BASE-VER: Bump to 7.0.1.
33318 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33320 * omp-low.c (omplow_simd_context): New struct. Use it...
33321 (lower_rec_simd_input_clauses): ...here and...
33322 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33323 references to idx, lane, max_vf, is_simt.
33325 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33327 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33330 2017-01-20 Martin Jambor <mjambor@suse.cz>
33332 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33333 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33335 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33336 (GTFILES): Rename hsa.c to hsa-common.c.
33337 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33338 * hsa-dump.c: Likewise.
33339 * hsa-gen.c: Likewise.
33340 * hsa-regalloc.c: Likewise.
33341 * ipa-hsa.c: Likewise.
33342 * omp-expand.c: Likewise.
33343 * omp-low.c: Likewise.
33344 * toplev.c: Likewise.
33346 2017-01-20 Marek Polacek <polacek@redhat.com>
33349 * doc/invoke.texi: Document -Wduplicated-branches.
33350 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33351 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33352 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33353 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33354 return 0 only when not OEP_LEXICOGRAPHIC.
33355 (fold_build_cleanup_point_expr): Use the expression
33356 location when building CLEANUP_POINT_EXPR.
33357 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33358 * tree.c (add_expr): Handle error_mark_node.
33360 2017-01-20 Martin Liska <mliska@suse.cz>
33363 * tree-profile.c (init_ic_make_global_vars): Do not call
33365 (gimple_init_gcov_profiler): Likewise.
33367 2017-01-20 Martin Liska <mliska@suse.cz>
33370 * cgraph.h (maybe_create_reference): Remove argument and
33372 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33374 * ipa-cp.c (create_specialized_node): Likewise.
33375 * symtab.c (symtab_node::maybe_create_reference): Handle
33376 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33378 2017-01-20 Martin Liska <mliska@suse.cz>
33380 * read-rtl-function.c (function_reader::create_function): Use
33381 build_decl instread of build_decl_stat.
33383 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33385 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33386 * config/i386/avx512dqintrin.h: Ditto.
33387 * config/i386/avx512fintrin.h: Ditto.
33388 * config/i386/i386-builtin-types.def: Add new types.
33389 * config/i386/i386.c: Handle new types.
33390 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33391 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33392 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33393 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33394 (__builtin_ia32_kshiftridi): New.
33395 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33397 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33401 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33402 define to rs6000_init_stack_protect_guard.
33403 (rs6000_init_stack_protect_guard): New function.
33405 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33406 Yunqiang Su <yunqiang.su@imgtec.com>
33408 * config.gcc (supported_defaults): Add madd4.
33409 (with_madd4): Add validation.
33410 (all_defaults): Add madd4.
33411 * config/mips/mips.opt (mmadd4): New option.
33412 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33414 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33416 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33417 (ISA_HAS_FUSED_MADD4): Likewise.
33418 * doc/invoke.texi (-mmadd4): Document the new option.
33419 * doc/install.texi (--with-madd4): Document the new option.
33421 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33423 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33424 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33425 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33426 (aarch64_init_pauth_hint_builtins): New.
33427 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33428 (aarch64_expand_builtin): Expand new builtins.
33430 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33432 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33433 * combine-stack-adj.c (no_unhandled_cfa): Handle
33434 REG_CFA_TOGGLE_RA_MANGLE.
33435 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33436 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33437 info for return address signing.
33438 (aarch64_expand_epilogue): Likewise.
33440 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33442 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33443 * config/aarch64/aarch64-protos.h
33444 (aarch64_return_address_signing_enabled): New declaration.
33445 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33447 (aarch64_expand_prologue): Sign return address before it's pushed onto
33449 (aarch64_expand_epilogue): Authenticate return address fetched from
33451 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33452 (aarch64_attributes): New function attributes for "sign-return-address".
33453 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33454 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33455 ("*do_return"): Generate combined instructions according to key index.
33456 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33457 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33459 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33460 * config/aarch64/aarch64.opt (msign-return-address=): New.
33461 * doc/extend.texi (AArch64 Function Attributes): Documents
33462 "sign-return-address=".
33463 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33465 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33467 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33468 overall option summary.
33470 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33472 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33473 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33474 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33475 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33477 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33479 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33480 -mpower9-minmax by default for -mcpu=power9.
33481 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33482 128-bit floating point.
33484 2017-01-20 Alan Modra <amodra@gmail.com>
33486 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33487 optimizing for size.
33489 2017-01-20 Alan Modra <amodra@gmail.com>
33492 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33493 for strcmp and strncmp from corresponding builtin decl.
33495 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33497 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33498 instead of i386/rtems-64.h.
33499 * config/i386/rtems-64.h: Remove.
33501 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33505 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33507 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33509 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33511 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33512 Change int to HOST_WIDE_INT.
33513 * config/aarch64/aarch64-protos.h
33514 (aarch64_simd_gen_const_vector_dup): Likewise.
33515 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33517 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33519 * langhooks-def.h (lhd_type_for_size): New decl.
33520 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33521 * langhooks.c (lhd_type_for_size): New function, taken from
33524 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33526 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33527 define_bypass for CR latency.
33528 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33529 (power9-alu2): Add define_bypass for CR latency.
33531 (power9-mul): Update insn latency.
33532 (power9-mul-compare): Update insn latency, bypass latency and remove
33535 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33537 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33539 * config/aarch64/aarch64.md
33540 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33541 aarch64_nopcrelative_literal_loads.
33542 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33544 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33546 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33547 TARGET_LOONGSON_3A.
33548 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33550 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33553 * config.gcc (supported_defaults): Add lxc1-sxc1.
33554 (with_lxc1_sxc1): Add validation.
33555 (all_defaults): Add lxc1-sxc1.
33556 * config/mips/mips.opt (mlxc1-sxc1): New option.
33557 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33559 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33560 __mips_no_lxc1_sxc1.
33561 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33562 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33563 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33565 2017-01-19 Richard Biener <rguenther@suse.de>
33567 PR tree-optimization/72488
33568 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33569 sure to restore SSA info.
33570 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33572 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33574 PR rtl-optimization/79121
33575 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33576 of the inner type when shifting an extended value.
33578 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33581 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33582 interposable aliases.
33584 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33587 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33588 Use the evmergelohi instruction.
33589 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33590 (mov_sitf_e500_subreg8_2_be): Likewise.
33591 (mov_sitf_e500_subreg12_2_le): Likewise.
33592 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33593 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33594 (mov_sitf_e500_subreg8_2_le): Likewise.
33595 (mov_sitf_e500_subreg12_2_be): Likewise.
33597 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33599 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33600 attribute from vecsimple to vecperm.
33601 (altivec_vbpermq2): Likewise.
33603 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33606 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33608 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33609 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33610 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33611 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33612 case where N arg is SIZE_MAX.
33613 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33614 (cmpstrsi): Add pattern.
33616 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33618 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33619 __builtin_vec_revb builtins.
33620 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33621 built-in functions to support generation of the ISA 3.0 XXBR<x>
33622 vector byte reverse instructions.
33623 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33624 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33625 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33626 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33627 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33628 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33629 (P9V_BUILTIN_VEC_REVB): Likewise.
33630 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33631 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33632 (p9_xxbrq_v16qi): Likewise.
33633 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33634 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33635 (p9_xxbrh_v8hi): Likewise.
33636 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33637 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33638 vec_revb built-in functions.
33640 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33642 PR rtl-optimization/78952
33643 * config/i386/i386.md (any_extract): New code iterator.
33644 (*insvqi_2): Use any_extract for source operand.
33645 (*insvqi_3): Use any_shiftrt for source operand.
33647 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33649 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33651 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33653 2017-01-18 Matthias Klose <doko@ubuntu.com>
33655 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33657 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33659 * config/rs6000/altivec.h (vec_bperm): Change #define.
33660 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33661 (altivec_vbpermq2): New define_insn.
33662 (altivec_vbpermd): Likewise.
33663 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33664 function interface.
33665 (VBPERMD): Likewise.
33666 (VBPERM): New polymorphic function interface.
33667 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33668 Add entries for P9V_BUILTIN_VEC_VBPERM.
33669 * doc/extend.texi: Add interfaces for vec_bperm.
33671 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33673 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33674 first letter of error messages.
33675 (s390_resolve_overloaded_builtin): Likewise.
33676 * config/s390/s390.c (s390_expand_builtin): Likewise.
33677 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33678 (s390_valid_target_attribute_inner_p): Likewise.
33679 * config/s390/s390.md ("tabort"): Likewise.
33681 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33683 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33684 (ISA_AVOID_DIV_HILO): New macro.
33685 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33686 (ISA_HAS_DDIV): Likewise.
33688 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33690 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33692 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33694 * doc/invoke.texi (fabi-version): Spelling fix.
33696 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33699 * doc/invoke.texi (fabi-version): Mention mangling fix for
33702 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33705 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33707 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33710 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33711 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33713 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33715 (stack_protect_test): Ditto.
33716 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33717 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33719 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33720 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33721 -mstack-protector-guard-offset=.
33722 (RS/6000 and PowerPC Options): Ditto.
33724 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33726 * config/i386/i386.h (MASK_CLASS_P): New define.
33727 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33728 there are no registers from different register sets also when
33729 mask registers are used. Update function comment.
33730 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33731 to (*k/*r) and (*k/*km) alternatives.
33733 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33735 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33736 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33737 (EH_RETURN_HANDLER_RTX): New define.
33738 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33739 Force frame pointer in EH return functions.
33740 (aarch64_expand_epilogue): Add barrier for eh_return.
33741 (aarch64_final_eh_return_addr): Remove.
33742 (aarch64_eh_return_handler_rtx): New function.
33743 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33745 (aarch64_eh_return_handler_rtx): New prototype.
33747 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33749 * config/rs6000/altivec.h (vec_rlmi): New #define.
33750 (vec_vrlnm): Likewise.
33751 (vec_rlnm): Likewise.
33752 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33753 (UNSPEC_VRLNM): Likewise.
33754 (VIlong): New mode iterator.
33755 (altivec_vrl<VI_char>mi): New define_insn.
33756 (altivec_vrl<VI_char>nm): Likewise.
33757 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33759 (VRLDNM): Likewise.
33760 (RLNM): New polymorphic function entry.
33761 (VRLWMI): New monomorphic function entry.
33762 (VRLDMI): Likewise.
33763 (RLMI): New polymorphic function entry.
33764 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33765 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33766 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33769 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33772 * dwarf2out.c (field_byte_offset): Restore the
33773 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33774 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33775 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33778 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33781 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33783 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33786 * dwarf2out.c (add_data_member_location_attribute): For constant
33787 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33788 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33789 DW_AT_byte_size attributes.
33791 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33793 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33794 after forcing to constant memory when the code model is medium.
33796 2017-01-17 Julia Koval <julia.koval@intel.com>
33799 * config/i386/avx512fintrin.h
33800 (_mm512_i32gather_ps): Change __addr type to void const*.
33801 (_mm512_mask_i32gather_ps): Ditto.
33802 (_mm512_i32gather_pd): Ditto.
33803 (_mm512_mask_i32gather_pd): Ditto.
33804 (_mm512_i64gather_ps): Ditto.
33805 (_mm512_mask_i64gather_ps): Ditto.
33806 (_mm512_i64gather_pd): Ditto.
33807 (_mm512_mask_i64gather_pd): Ditto.
33808 (_mm512_i32gather_epi32): Ditto.
33809 (_mm512_mask_i32gather_epi32): Ditto.
33810 (_mm512_i32gather_epi64): Ditto.
33811 (_mm512_mask_i32gather_epi64): Ditto.
33812 (_mm512_i64gather_epi32): Ditto.
33813 (_mm512_mask_i64gather_epi32): Ditto.
33814 (_mm512_i64gather_epi64): Ditto.
33815 (_mm512_mask_i64gather_epi64): Ditto.
33816 (_mm512_i32scatter_ps): Change __addr type to void*.
33817 (_mm512_mask_i32scatter_ps): Ditto.
33818 (_mm512_i32scatter_pd): Ditto.
33819 (_mm512_mask_i32scatter_pd): Ditto.
33820 (_mm512_i64scatter_ps): Ditto.
33821 (_mm512_mask_i64scatter_ps): Ditto.
33822 (_mm512_i64scatter_pd): Ditto.
33823 (_mm512_mask_i64scatter_pd): Ditto.
33824 (_mm512_i32scatter_epi32): Ditto.
33825 (_mm512_mask_i32scatter_epi32): Ditto.
33826 (_mm512_i32scatter_epi64): Ditto.
33827 (_mm512_mask_i32scatter_epi64): Ditto.
33828 (_mm512_i64scatter_epi32): Ditto.
33829 (_mm512_mask_i64scatter_epi32): Ditto.
33830 (_mm512_i64scatter_epi64): Ditto.
33831 (_mm512_mask_i64scatter_epi64): Ditto.
33832 * config/i386/avx512pfintrin.h
33833 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33834 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33835 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33836 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33837 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33838 (_mm512_prefetch_i32scatter_ps): Ditto.
33839 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33840 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33841 (_mm512_prefetch_i64scatter_pd): Ditto.
33842 (_mm512_prefetch_i64scatter_ps): Ditto.
33843 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33844 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33845 * config/i386/avx512vlintrin.h
33846 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33847 (_mm_mmask_i32gather_ps): Ditto.
33848 (_mm256_mmask_i32gather_pd): Ditto.
33849 (_mm_mmask_i32gather_pd): Ditto.
33850 (_mm256_mmask_i64gather_ps): Ditto.
33851 (_mm_mmask_i64gather_ps): Ditto.
33852 (_mm256_mmask_i64gather_pd): Ditto.
33853 (_mm_mmask_i64gather_pd): Ditto.
33854 (_mm256_mmask_i32gather_epi32): Ditto.
33855 (_mm_mmask_i32gather_epi32): Ditto.
33856 (_mm256_mmask_i32gather_epi64): Ditto.
33857 (_mm_mmask_i32gather_epi64): Ditto.
33858 (_mm256_mmask_i64gather_epi32): Ditto.
33859 (_mm_mmask_i64gather_epi32): Ditto.
33860 (_mm256_mmask_i64gather_epi64): Ditto.
33861 (_mm_mmask_i64gather_epi64): Ditto.
33862 (_mm256_i32scatter_ps): Change __addr type to void*.
33863 (_mm256_mask_i32scatter_ps): Ditto.
33864 (_mm_i32scatter_ps): Ditto.
33865 (_mm_mask_i32scatter_ps): Ditto.
33866 (_mm256_i32scatter_pd): Ditto.
33867 (_mm256_mask_i32scatter_pd): Ditto.
33868 (_mm_i32scatter_pd): Ditto.
33869 (_mm_mask_i32scatter_pd): Ditto.
33870 (_mm256_i64scatter_ps): Ditto.
33871 (_mm256_mask_i64scatter_ps): Ditto.
33872 (_mm_i64scatter_ps): Ditto.
33873 (_mm_mask_i64scatter_ps): Ditto.
33874 (_mm256_i64scatter_pd): Ditto.
33875 (_mm256_mask_i64scatter_pd): Ditto.
33876 (_mm_i64scatter_pd): Ditto.
33877 (_mm_mask_i64scatter_pd): Ditto.
33878 (_mm256_i32scatter_epi32): Ditto.
33879 (_mm256_mask_i32scatter_epi32): Ditto.
33880 (_mm_i32scatter_epi32): Ditto.
33881 (_mm_mask_i32scatter_epi32): Ditto.
33882 (_mm256_i32scatter_epi64): Ditto.
33883 (_mm256_mask_i32scatter_epi64): Ditto.
33884 (_mm_i32scatter_epi64): Ditto.
33885 (_mm_mask_i32scatter_epi64): Ditto.
33886 (_mm256_i64scatter_epi32): Ditto.
33887 (_mm256_mask_i64scatter_epi32): Ditto.
33888 (_mm_i64scatter_epi32): Ditto.
33889 (_mm_mask_i64scatter_epi32): Ditto.
33890 (_mm256_i64scatter_epi64): Ditto.
33891 (_mm256_mask_i64scatter_epi64): Ditto.
33892 (_mm_i64scatter_epi64): Ditto.
33893 (_mm_mask_i64scatter_epi64): Ditto.
33894 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33895 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33896 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33897 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33898 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33899 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33900 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33901 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33902 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33903 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33904 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33905 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33906 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33907 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33908 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33909 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33910 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33911 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33912 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33913 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33914 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33915 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33916 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33917 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33918 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33919 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33920 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33921 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33922 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33923 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33924 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33925 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33926 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33927 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33928 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33929 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33930 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33931 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33932 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33933 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33934 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33935 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33936 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33937 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33938 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33939 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33940 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33941 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33942 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33943 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33944 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33945 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33946 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33947 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33948 definitions accordingly.
33950 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33951 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33954 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33957 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33960 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33961 hard regs for inner regno.
33963 2017-01-17 Martin Liska <mliska@suse.cz>
33966 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33967 assumption and add comment.
33969 2017-01-17 Nathan Sidwell <nathan@acm.org>
33971 * ipa-visibility.c (localize_node): New function, broken out of ...
33972 (function_and_variable_visibility): ... here. Call it.
33974 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33976 PR middle-end/77445
33977 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33978 correctly set frequency of oudgoing edge.
33979 (duplicate_thread_path): Fix profile updating.
33981 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33984 * configure.ac: Add GCC_BASE_VER.
33985 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33986 version from BASE-VER file.
33987 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33988 (gcc.o): Depend on $(BASEVER).
33989 * common.opt (dumpfullversion): New option.
33990 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33991 * doc/invoke.texi: Document -dumpfullversion.
33992 * doc/install.texi: Document --with-gcc-major-version-only.
33993 * configure: Regenerated.
33995 2017-01-17 Richard Biener <rguenther@suse.de>
33997 PR tree-optimization/71433
33998 * tree-vrp.c (register_new_assert_for): Merge same asserts
33999 on all incoming edges.
34000 (process_assert_insertions_for): Handle insertions at the
34003 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34005 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34006 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34008 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34011 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34014 2017-01-17 Alan Modra <amodra@gmail.com>
34017 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34018 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34019 symbolic stack limit when pic.
34021 2017-01-16 Martin Sebor <msebor@redhat.com>
34023 PR tree-optimization/78608
34024 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34026 2017-01-16 Jeff Law <law@redhat.com>
34029 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34030 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34031 for several include directories that may be relative to sysroot.
34032 * config/i386/x-mingw32 (gplus_includedir): Define.
34033 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34034 (native_system_includedir): Likewise.
34035 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34036 override if TARGET_SYSTEM_ROOT is defined.
34037 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34039 PR tree-optimization/79090
34040 PR tree-optimization/33562
34041 PR tree-optimization/61912
34042 PR tree-optimization/77485
34043 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34044 and computed trims into the dump file.
34046 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34048 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34050 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34053 * gimplify.c (gimplify_init_constructor): If want_value and
34054 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34058 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34059 sequence. Formatting fixes.
34060 (doloop_optimize): Formatting fixes.
34063 * gcc.c (debug_level_greater_than_spec_func): New function.
34064 (static_spec_functions): Add debug-level-gt spec function.
34065 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34067 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34068 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34069 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34070 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34071 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34072 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34074 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34076 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34077 QImode fixups to general and mask registers only.
34079 2017-01-16 Carl Love <cel@us.ibm.com>
34081 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34082 for built-in functions
34083 vector signed char vec_nabs (vector signed char)
34084 vector signed short vec_nabs (vector signed short)
34085 vector signed int vec_nabs (vector signed int)
34086 vector signed long long vec_nabs (vector signed long long)
34087 vector float vec_nabs (vector float)
34088 vector double vec_nabs (vector double)
34089 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34091 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34092 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34093 * doc/extend.texi: Update the documentation file for the new built-in
34096 2017-01-16 Martin Sebor <msebor@redhat.com>
34098 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34101 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34103 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34104 UNSPEC_VSX__XXSPLTD to require special splat handling.
34106 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34109 * system.h: Poison strndup.
34111 2017-01-16 Alan Modra <amodra@gmail.com>
34114 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34117 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34119 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34121 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34123 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34124 call recog here. Assert that INSN_CODE (insn) is non-negative.
34126 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34129 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34131 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34132 in the currently scheduled RTL fragment.
34134 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34136 PR rtl-optimization/78751
34137 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34140 2017-01-14 Jeff Law <law@redhat.com>
34142 PR tree-optimization/79090
34143 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34144 variable length stores.
34145 (compute_trims): Delete dead assignment to *trim_tail.
34146 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34149 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34151 PR rtl-optimization/78626
34152 PR rtl-optimization/78727
34153 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34154 of a block, and split such blocks after everything else is finished.
34156 2017-01-14 Alan Modra <amodra@gmail.com>
34159 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34160 target legitimate_combined_insn.
34161 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34162 (rs6000_legitimate_combined_insn): New function.
34163 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34165 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34166 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34167 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34169 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34171 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34173 2017-01-13 Jeff Law <law@redhat.com>
34175 PR tree-optimization/33562
34176 PR tree-optimization/61912
34177 PR tree-optimization/77485
34178 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34180 (delete_dead_assignment): Likewise.
34181 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34182 statement to delete_dead_call and delete_dead_assignment.
34184 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34187 * substring-locations.c (format_warning_va): Strengthen case 1 so
34188 that both endpoints of the substring must be within the format
34189 range for just the substring to be printed.
34191 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34193 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34194 * config/i386/i386.c (ix86_target_string): Add missing options
34195 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34196 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34197 flags_other and ix86_target_other to flags2_other. Display unknown
34199 (ix86_valid_target_attribute_inner_p): Add missing options and
34200 reorder options by implied ISAs, as in ix86_target_string.
34202 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34204 * hash-table.h (hash_table::too_empty_p): New function.
34205 (hash_table::expand): Use it.
34206 (hash_table::traverse): Likewise.
34207 (hash_table::empty_slot): Use sizeof (value_type) instead of
34208 sizeof (PTR) to convert bytes to elements. Shrink the table
34209 if the current size is excessive for the current number of
34212 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34214 * ira-costs.c (record_reg_classes): Break from the inner loop
34215 early once alt_fail is known to be true. Update outer loop
34216 handling accordingly.
34218 2017-01-13 Jeff Law <law@redhat.com>
34220 * tree-ssa-dse.c (decrement_count): New function.
34221 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34222 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34223 when we know the partially dead statement is a mem* function.
34225 PR tree-optimization/61912
34226 PR tree-optimization/77485
34227 * tree-ssa-dse.c: Include expr.h.
34228 (maybe_trim_constructor_store): New function.
34229 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34231 PR tree-optimization/33562
34232 PR tree-optimization/61912
34233 PR tree-optimization/77485
34234 * doc/invoke.texi: Document new dse-max-object-size param.
34235 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34236 * tree-ssa-dse.c: Include params.h.
34237 (dse_store_status): New enum.
34238 (initialize_ao_ref_for_dse): New, partially extracted from
34240 (valid_ao_ref_for_dse, normalize_ref): New.
34241 (setup_live_bytes_from_ref, compute_trims): Likewise.
34242 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34243 (maybe_trim_partially_dead_store): Likewise.
34244 (maybe_trim_complex_store): Likewise.
34245 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34246 Track what bytes live from the original store. Return tri-state
34247 for dead, partially dead or live.
34248 (dse_dom_walker): Add constructor, destructor and new private members.
34249 (delete_dead_call, delete_dead_assignment): New extracted from
34251 (dse_optimize_stmt): Make a member of dse_dom_walker.
34252 Use initialize_ao_ref_for_dse.
34254 PR tree-optimization/33562
34255 PR tree-optimization/61912
34256 PR tree-optimization/77485
34257 * sbitmap.h (bitmap_count_bits): Prototype.
34258 (bitmap_clear_range, bitmap_set_range): Likewise.
34259 * sbitmap.c (bitmap_clear_range): New function.
34260 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34262 2017-01-13 Martin Liska <mliska@suse.cz>
34265 * function.c (set_cfun): Add new argument force.
34266 * function.h (set_cfun): Likewise.
34267 * ipa-inline-transform.c (inline_call): Use the function when
34268 strict alising from is dropped for function we inline to.
34270 2017-01-13 Richard Biener <rguenther@suse.de>
34272 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34273 for dumping GIMPLE INTEGER_CSTs.
34275 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34277 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34278 to 201112L since C++17.
34280 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34283 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34284 if -fsanitize=kernel-address is present.
34286 2017-01-13 Richard Biener <rguenther@suse.de>
34288 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34289 as _Literal ( type ) number in case usual suffixes do not
34290 preserve all information.
34292 2017-01-13 Richard Biener <rguenther@suse.de>
34294 PR tree-optimization/77283
34295 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34296 and ssa-iterators.h.
34297 (is_feasible_trace): Implement a cost model based on joiner
34300 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34303 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34304 char or short to __float128/_Float128 directly.
34306 2017-01-12 Martin Sebor <msebor@redhat.com>
34308 to -Wformat-overflow.
34309 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34310 (min_bytes_remaining): Same.
34311 (get_string_length): Same.
34312 (format_string): Same.
34313 (format_directive): Same.
34315 (pass_sprintf_length::handle_gimple_call): Same.
34317 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34319 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34320 info.nowrite calls with no lhs that can't throw. Return bool
34321 whether gsi_remove has been called or not.
34322 (pass_sprintf_length::handle_gimple_call): Return bool whether
34323 try_substitute_return_value called gsi_remove. Formatting fix.
34324 (pass_sprintf_length::execute): Don't use gsi_remove if
34325 handle_gimple_call returned true.
34328 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34329 be removed due to side-effects, don't remove following barrier nor
34330 turn the successor edge into fallthru edge.
34332 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34335 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34336 element-reversing loads and stores as not swappable.
34338 2017-01-12 Nathan Sidwell <nathan@acm.org>
34339 Nicolai Stange <nicstange@gmail.com>
34341 * combine.c (try_combine): Don't ignore result of overlap checking
34342 loop. Combine overlap & asm check into single loop.
34344 2017-01-12 Richard Biener <rguenther@suse.de>
34346 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34347 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34349 2017-01-12 Richard Biener <rguenther@suse.de>
34351 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34352 and TS_TARGET_OPTION directly derive from TS_BASE.
34353 * tree-core.h (tree_optimization_option): Derive from tree_base.
34354 (tree_target_option): Likewise.
34356 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34358 * config/i386/i386.c (memory_address_length): Increase len
34359 only when rip_relative_addr_p returns false.
34361 2017-01-11 Julia Koval <julia.koval@intel.com>
34363 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34364 (OPTION_MASK_ISA_SGX_SET): New.
34365 (ix86_handle_option): Handle OPT_msgx.
34366 * config.gcc: Added sgxintrin.h.
34367 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34368 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34369 * config/i386/i386.c (ix86_target_string): Add -msgx.
34371 (ix86_option_override_internal): Handle new options.
34372 (ix86_valid_target_attribute_inner_p): Add sgx.
34373 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34374 * config/i386/i386.opt: Add msgx.
34375 * config/i386/sgxintrin.h: New file.
34376 * config/i386/x86intrin.h: Add sgxintrin.h.
34378 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34381 * fold-const.c (maybe_nonzero_address): Return 1 for function
34383 (tree_single_nonzero_warnv_p): Don't handle function local objects
34387 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34390 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34393 * opts.c: Include "spellcheck.h"
34394 (struct string_fragment): New struct.
34395 (struct edit_distance_traits<const string_fragment &>): New
34397 (get_closest_sanitizer_option): New function.
34398 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34400 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34402 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34404 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34405 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34406 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34407 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34408 for initial die_offset if dwarf_split_debug_info.
34409 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34410 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34412 (output_skeleton_debug_sections): Formatting fix. Use
34413 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34414 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34416 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34418 * config/arm/cortex-a53.md: Add bypasses for
34419 cortex_a53_r2f_cvt.
34420 (cortex_a53_r2f): Only use for transfers.
34421 (cortex_a53_f2r): Likewise.
34422 (cortex_a53_r2f_cvt): Add reservation for conversions.
34423 (cortex_a53_f2r_cvt): Likewise.
34425 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34427 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34428 to all inlined functions, change static to extern.
34430 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34433 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34435 (arm_assemble_integer): Likewise.
34437 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34439 * config.gcc: Use new awk script to check CPU, FPU and architecture
34440 parameters for --with-... options.
34441 * config/arm/parsecpu.awk: New file
34442 * config/arm/arm-cpus.in: New file.
34443 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34445 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34447 * config/arm/t-arm: Update dependency rules.
34448 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34449 of processing .def files.
34450 * config/arm/genopt.sh: Deleted.
34451 * config/arm/gentune.sh: Deleted.
34452 * config/arm/arm-cores.def: Deleted.
34453 * config/arm/arm-arches.def: Deleted.
34454 * config/arm/arm-fpus.def: Deleted.
34455 * config/arm/arm-tune.md: Regenerated.
34456 * config/arm/arm-tables.opt: Regenerated.
34457 * config/arm/arm-cpu.h: New generated file.
34458 * config/arm/arm-cpu-data.h: New generated file.
34459 * config/arm/arm-cpu-cdata.h: New generated file.
34461 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34464 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34466 (input_varpool_node): Unpack dynamically_initialized bit.
34468 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34470 PR rtl-optimization/79032
34471 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34472 the alignment of the adjusted memory reference against that of MODE,
34473 instead of the alignment of the original memory reference.
34475 2017-01-11 Martin Jambor <mjambor@suse.cz>
34477 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34479 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34480 decorated functions.
34482 2017-01-11 Richard Biener <rguenther@suse.de>
34484 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34485 set range/nonnull info for PHI results. Do not set it on
34486 stmts marked for removal.
34488 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34490 * expr.c (store_field): In the bitfield case, fetch the return value
34491 from the registers before applying a single big-endian adjustment.
34492 Always do a final load for a BLKmode value not larger than a word.
34494 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34497 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34498 that we correctly handle column numbers greater than
34499 LINE_MAP_MAX_COLUMN_NUMBER.
34501 2017-01-10 Martin Sebor <msebor@redhat.com>
34503 PR middle-end/78245
34504 * gimple-ssa-sprintf.c (get_destination_size): Call
34505 {init,fini}object_sizes.
34506 * tree-object-size.c (addr_object_size): Adjust.
34507 (pass_through_call): Adjust.
34508 (pass_object_sizes::execute): Adjust.
34509 * tree-object-size.h (fini_object_sizes): Declare.
34511 2017-01-10 Martin Sebor <msebor@redhat.com>
34513 PR tree-optimization/78775
34514 * builtins.c (get_size_range): Move...
34515 * calls.c: ...to here.
34516 (alloc_max_size): Accept zero argument.
34517 (operand_signed_p): Remove.
34518 (maybe_warn_alloc_args_overflow): Call get_size_range.
34519 * calls.h (get_size_range): Declare.
34521 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34523 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34524 from TI's devices.csv file as of September 2016.
34525 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34527 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34529 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34530 * doc/invoke.texi: Likewise.
34531 * doc/md.texi: Likewise.
34532 * doc/objc.texi: Likewise.
34534 2017-01-10 Joshua Conner <joshconner@google.com>
34536 * config/arm/fuchsia-elf.h: New file.
34537 * config/fuchsia.h: New file.
34538 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34539 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34541 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34543 2016-01-10 Richard Biener <rguenther@suse.de>
34545 PR tree-optimization/79034
34546 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34547 Propagate out degenerate PHIs in the joiner.
34549 2017-01-10 Martin Liska <mliska@suse.cz>
34551 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34552 (sort_congruence_classes_by_decl_uid): Likewise.
34553 (sort_congruence_class_groups_by_decl_uid): Likewise.
34554 (sem_item_optimizer::merge_classes): Sort class, groups in these
34555 classes and members in the groups by DECL_UID of declarations.
34556 This would make merge operations stable.
34558 2017-01-10 Martin Liska <mliska@suse.cz>
34560 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34561 usage of m_classes_vec.
34562 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34563 (sem_item_optimizer::get_group_by_hash): Likewise.
34564 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34565 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34566 (sem_item_optimizer::verify_classes): Likewise.
34567 (sem_item_optimizer::process_cong_reduction): Likewise.
34568 (sem_item_optimizer::dump_cong_classes): Likewise.
34569 (sem_item_optimizer::merge_classes): Likewise.
34570 * ipa-icf.h (congruence_class_hash): Rename from
34571 congruence_class_group_hash. Remove declaration of m_classes_vec.
34573 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34575 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34576 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34577 * config.gcc: Add avx512vpopcntdqintrin.h.
34578 * config/i386/avx512vpopcntdqintrin.h: New.
34579 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34580 * config/i386/i386-builtin-types.def: Add new types.
34581 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34582 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34583 __builtin_ia32_vpopcountq_v8di_mask): New.
34584 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34585 __AVX512VPOPCNTDQ__.
34586 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34587 (PTA_AVX512VPOPCNTDQ): Define.
34588 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34589 TARGET_AVX512VPOPCNTDQ_P): Define.
34590 * config/i386/i386.opt: Add mavx512vpopcntdq.
34591 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34592 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34594 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34596 PR middle-end/77484
34597 * predict.def (PRED_CALL): Set to 67.
34599 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34601 * expr.c (store_field): In the bitfield case, if the value comes from
34602 a function call and is of an aggregate type returned in registers, do
34603 not modify the field mode; extract the value in all cases if the mode
34604 is BLKmode and the size is not larger than a word.
34606 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34609 * config/i386/cpuid.h: Fix undefined behavior.
34611 2017-01-04 Jeff Law <law@redhat.com>
34613 PR tree-optimization/79007
34614 PR tree-optimization/67955
34615 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34616 conservative for pt.null when flag_non_call_exceptions is on.
34618 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34620 PR translation/79019
34621 PR translation/79020
34622 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34623 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34624 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34626 * config/avr/avr.opt (maccumulate-args): Likewise.
34627 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34628 * common.opt (freport-bug): Likewise.
34629 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34630 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34631 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34632 translatable string.
34633 * config/i386/i386.c (function_value_32): Likewise.
34634 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34635 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34637 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34638 * common/config/msp430/msp430-common.c (msp430_handle_option):
34640 * symtab.c (symtab_node::verify_base): Likewise.
34641 * opts.c (set_debug_level): Likewise.
34642 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34643 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34644 missing whitespace to translatable strings.
34645 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34646 * config/sh/superh.h: Likewise.
34647 * config/i386/xopintrin.h: Likewise.
34648 * config/i386/znver1.md: Likewise.
34649 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34650 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34651 * double-int.h (struct double_int): Likewise.
34652 * double-int.c (div_and_round_double): Likewise.
34653 * wide-int.cc: Likewise.
34654 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34655 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34656 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34657 (crossjumps_occurred): ... this.
34658 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34661 PR tree-optimization/78899
34662 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34663 returning bool return struct loop *, NULL for failure and the new
34665 (versionable_outer_loop_p): Don't version outer loop if it has
34666 dont_vectorized bit set.
34667 (tree_if_conversion): When versioning outer loop, ensure
34668 tree_if_conversion is performed also on the inner loop of the
34669 non-vectorizable outer loop copy.
34670 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34671 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34672 prevent vectorization of it.
34673 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34674 the outer loop vectorization of the non-scalar version is attempted
34675 before vectorization of the inner loop in scalar version. If
34676 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34677 vectorization of its inner loop.
34678 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34679 has 2 inner loops, rename also on edges from bb whose single pred
34680 is outer_loop->header. Fix typo in function comment.
34682 2017-01-09 Martin Sebor <msebor@redhat.com>
34685 * asan.c (asan_emit_stack_protection): Increase local buffer size
34686 to avoid snprintf truncation warning.
34688 2017-01-09 Andrew Pinski <apinski@cavium.com>
34690 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34691 to reference thunderx2t99 for the tuning structure
34692 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34694 (thunderx2t99_extra_costs): This.
34695 * config/aarch64/aarch64-tune.md: Regenerate.
34696 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34697 (vulcan_addrcost_table): This.
34698 (vulcan_regmove_cost): Rename to ...
34699 (thunderx2t99_regmove_cost): This.
34700 (vulcan_vector_cost): Rename to ...
34701 (thunderx2t99_vector_cost): this.
34702 (vulcan_branch_cost): Rename to ...
34703 (thunderx2t99_branch_cost): This.
34704 (vulcan_tunings): Rename to ...
34705 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34706 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34708 2017-01-09 Martin Jambor <mjambor@suse.cz>
34712 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34713 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34714 (propagate_vr_accross_jump_function): Use the above function for all
34715 value range computations for pass-through jump functions and type
34716 converasion from explicit value range values.
34717 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34718 parameters from TYPE_ARG_TYPES.
34719 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34720 (ipa_write_node_info): Stream type of the actual argument.
34721 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34723 2017-01-09 Martin Liska <mliska@suse.cz>
34726 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34727 (lookup_compiler): Do not show error message with have_E.
34729 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34731 PR tree-optimization/78938
34732 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34733 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34734 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34735 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34738 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34740 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34743 2017-01-09 Richard Biener <rguenther@suse.de>
34745 PR tree-optimization/78997
34746 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34747 name condition properly.
34749 2017-01-09 Richard Biener <rguenther@suse.de>
34752 * dwarf2out.c (is_cxx): New overload with context.
34753 (is_naming_typedef_decl): Use it.
34755 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34757 * invoke.texi (Option Summary): Correct spacing in option lists
34758 and add line breaks to fix over-long lines.
34760 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34762 PR middle-end/17660
34764 * extend.texi (Common Variable Attributes): Add xref to GCC
34765 Internals manual to explain mode attribute keywords.
34767 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34770 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34771 and Preprocessor Options.
34772 (Options for Linking): Document -pthread here....
34773 (RS/6000 and PowerPC Options): ...not here.
34774 (Solaris 2 Options): ...or here.
34775 * doc/cppopts.texi: Document -pthread.
34777 2017-01-08 Martin Sebor <msebor@redhat.com>
34779 PR middle-end/77708
34780 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34781 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34782 New member functions.
34783 (format_directive): Used them.
34785 (pass_sprintf_length::handle_gimple_call): Same.
34786 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34787 to avoid truncation for any argument.
34788 (extract_affine_mul): Same.
34789 * tree.c (get_file_function_name): Same.
34791 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34793 PR middle-end/77484
34794 * predict.def (PRED_INDIR_CALL): Set to 86.
34796 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34798 PR preprocessor/54124
34799 * doc/cppopts.texi: Reformat -d subtable to list the full name
34800 of the options. Add cross-reference to the docs for the general
34801 compiler -d options.
34802 * doc/invoke.texi (Developer Options): Add cross-reference to the
34803 preprocessor-specific -d option documentation.
34805 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34807 PR preprocessor/13498
34808 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34809 redudant material, and reflect new command-line options.
34810 (System Headers): Likewise.
34812 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34814 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34815 -isystem, and -idirafter. Copy-edit.
34816 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34817 default for -ftrack-macro-expansion. Delete obsolete and
34818 badly-formatted implementation details about -fdebug-cpp output.
34819 * doc/cppwarnopts.texi: Copy-edit.
34821 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34824 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34825 that the transition from a max line width >= 1<<10 to narrower
34826 lines works correctly.
34828 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34830 * doc/options.texi (PerFunction): New.
34831 * opt-functions.awk (switch_flags): Map both Optimization and
34832 PerFunction to CL_OPTIMIZATION.
34833 * opth-gen.awk: Test for PerFunction flag along with
34835 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34836 it only when the latter is present. Skip those that don't in
34837 the hash function generator.
34838 * common.opt (fvar-tracking): Mark as PerFunction instead of
34840 (fvar-tracking-assignments): Likewise.
34841 (fvar-tracking-assignments-toggle): Likewise.
34842 (fvar-tracking-uninit): Likewise.
34844 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34846 PR translation/79018
34847 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34850 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34853 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34854 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34855 TARGET_LONG_JUMP_TABLE_OFFSETS.
34856 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34857 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34858 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34859 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34860 * config/m68k/m68k.md (tablejump expander): Likewise.
34861 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34862 TARGET_LONG_JUMP_TABLE_OFFSETS.
34863 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34864 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34866 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34867 David Holsgrove <david.holsgrove@xilinx.com>
34869 * common/config/microblaze/microblaze-common.c
34870 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34871 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34873 * config/microblaze/microblaze.c (microblaze_must_save_register)
34874 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34876 (microblaze_eh_return): New function.
34877 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34878 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34879 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34880 * config/microblaze/microblaze.md (eh_return): New pattern.
34882 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34884 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34885 GCC_DIAGNOSTIC_STRINGIFY): Define.
34887 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34889 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34891 * config/arm/arm.md (<mcrr>): New.
34893 * config/arm/arm.c (arm_arch5te): New.
34894 (arm_option_override): Set arm_arch5te.
34895 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34897 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34898 (arm_mcrr_qualifiers): ... this. New.
34899 (MRRC_QUALIFIERS): Define to...
34900 (arm_mrrc_qualifiers): ... this. New.
34901 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34903 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34904 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34905 (MRRCI, mrrc, MRRC): New.
34906 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34907 VUNSPEC_MRRC2): New.
34909 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34911 * config/arm/arm.md (<mcr>): New.
34913 * config/arm/arm.c (arm_coproc_builtin_available): Add
34914 support for mcr, mrc, mcr2 and mrc2.
34915 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34916 (arm_mcr_qualifiers): ... this. New.
34917 (MRC_QUALIFIERS): Define to ...
34918 (arm_mrc_qualifiers): ... this. New.
34919 (MCR_QUALIFIERS): Define to ...
34920 (arm_mcr_qualifiers): ... this. New.
34921 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34923 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34924 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34925 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34926 VUNSPEC_MRC2): New.
34928 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34930 * config/arm/arm.md (*ldc): New.
34934 * config/arm/arm.c (arm_coproc_builtin_available): Add
34935 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34936 (arm_coproc_ldc_stc_legitimate_address): New.
34937 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34938 'qualifier_const_pointer'.
34939 (LDC_QUALIFIERS): Define to...
34940 (arm_ldc_qualifiers): ... this. New.
34941 (STC_QUALIFIERS): Define to...
34942 (arm_stc_qualifiers): ... this. New.
34943 * config/arm/arm-protos.h
34944 (arm_coproc_ldc_stc_legitimate_address): New.
34945 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34946 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34947 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34948 stc2, stcl, stc2l): New.
34949 * config/arm/constraints.md (Uz): New.
34950 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34951 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34952 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34953 VUNSPEC_STC2L): New.
34955 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34957 * config/arm/arm.md (<cdp>): New.
34958 * config/arm/arm.c (neon_const_bounds): Rename this ...
34959 (arm_const_bounds): ... this.
34960 (arm_coproc_builtin_available): New.
34961 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34962 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34963 (CDP_QUALIFIERS): Define to...
34964 (arm_cdp_qualifiers): ... this. New.
34966 (arm_expand_builtin_args): Add case for 6 arguments.
34967 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34968 (arm_const_bounds): ... this.
34969 (arm_coproc_builtin_available): New.
34970 * config/arm/arm_acle.h (__arm_cdp): New.
34972 * config/arm/arm_acle_builtins.def (cdp): New.
34974 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34975 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34976 'arm_const_bounds'.
34977 * config/arm/types.md (coproc): New.
34978 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34979 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34980 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34981 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34983 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34985 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34986 (UBINOP_QUALIFIERS): New.
34988 (acle_builtin_data): New. Change comment.
34989 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34990 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34991 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34992 arm_acle_builtins.def.
34993 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34994 (arm_init_acle_builtins): New.
34995 (CRC32_BUILTIN): Remove.
34996 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34997 crc32cb, crc32ch and crc32cw.
34998 (arm_init_crc32_builtins): Remove.
34999 (arm_init_builtins): Use arm_init_acle_builtins rather
35000 than arm_init_crc32_builtins.
35001 (arm_expand_acle_builtin): New.
35002 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35003 * config/arm/arm_acle_builtins.def: New.
35005 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35007 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35008 (arm_builtin_datum): ... this.
35009 (arm_init_neon_builtin): Rename to ...
35010 (arm_init_builtin): ... this. Add a new parameters PREFIX
35011 and USE_SIG_IN_NAME.
35012 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35013 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35014 'arm_builtin_datum'.
35015 (arm_init_vfp_builtins): Likewise.
35016 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35017 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35018 (arm_expand_neon_args): Rename to ...
35019 (arm_expand_builtin_args): ... this. Rename builtin_arg
35020 enum values and differentiate between ARG_BUILTIN_MEMORY
35021 and ARG_BUILTIN_NEON_MEMORY.
35022 (arm_expand_neon_builtin_1): Rename to ...
35023 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35024 values, arm_expand_builtin_args and add bool parameter NEON.
35025 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35026 (arm_expand_vfp_builtin): Likewise.
35027 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35029 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35031 PR middle-end/77484
35032 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35033 * predict.c (tree_estimate_probability_bb): Reverse direction of
35034 polymorphic call predictor.
35036 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35038 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35039 (determine_pass_name_match): ...this new function and...
35040 (should_skip_pass_p): ...this new function.
35042 2017-01-06 Nathan Sidwell <nathan@acm.org>
35044 * ipa-visibility.c (function_and_variable_visibility): Reformat
35045 comments and long lines. Remove extrneous if.
35046 * symtab.c (symtab_node::make_decl_local): Fix code format.
35047 (symtab_node::set_section_for_node): Fix comment typo.
35049 2017-01-06 Martin Liska <mliska@suse.cz>
35052 * lra-constraints.c: Rename invariant to lra_invariant.
35053 * predict.c (set_even_probabilities): Initialize e to NULL.
35055 2017-01-05 Martin Sebor <msebor@redhat.com>
35057 PR tree-optimization/78910
35058 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35059 (format_integer): Correct off-by-one error in the handling
35060 of precision with negative numbers in signed conversions..
35062 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35064 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35066 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35068 PR tree-optimization/71016
35069 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35070 factor_out_conditional_conversion. Formatting fix.
35071 (factor_out_conditional_conversion): Add cond_stmt argument.
35072 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35073 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35076 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35078 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35079 read-rtl-function.o, and selftest-rtl.o.
35080 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35081 (selftest::aarch64_test_loading_full_dump): New function.
35082 (selftest::aarch64_run_selftests): New function.
35083 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35084 selftest::aarch64_run_selftests.
35085 * config/i386/i386.c
35086 (selftest::ix86_test_loading_dump_fragment_1): New function.
35087 (selftest::ix86_test_loading_call_insn): New function.
35088 (selftest::ix86_test_loading_full_dump): New function.
35089 (selftest::ix86_test_loading_unspec): New function.
35090 (selftest::ix86_run_selftests): Call the new functions.
35091 * emit-rtl.c (maybe_set_max_label_num): New function.
35092 * emit-rtl.h (maybe_set_max_label_num): New decl.
35093 * function.c (instantiate_decls): Guard call to
35094 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35095 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35097 * gensupport.c (gen_reader::gen_reader): Pass "false"
35098 for new "compact" param of rtx_reader.
35099 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35100 rather than an empty string for NULL strings.
35101 * read-md.c: Potentially include config.h rather than bconfig.h.
35102 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35103 (have_error): New global, copied from errors.c.
35104 (md_reader::read_name): Rename to...
35105 (md_reader::read_name_1): ...this, adding "out_loc" param,
35106 and converting "missing name or number" to returning false, rather
35108 (md_reader::read_name): Reimplement in terms of read_name_1.
35109 (md_reader::read_name_or_nil): New function.
35110 (md_reader::read_string): Handle "(nil)" by returning NULL.
35111 (md_reader::md_reader): Add new param "compact".
35112 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35113 (md_reader::read_file): New method.
35114 * read-md.h (md_reader::md_reader): Add new param "compact".
35115 (md_reader::read_file): New method.
35116 (md_reader::is_compact): New accessor.
35117 (md_reader::read_name): Convert return type from void to file_location.
35118 (md_reader::read_name_or_nil): New decl.
35119 (md_reader::read_name_1): New decl.
35120 (md_reader::m_compact): New field.
35121 (noop_reader::noop_reader): Pass "false" for new "compact" param
35123 (rtx_reader::rtx_reader): Add new "compact" param.
35124 (rtx_reader::read_rtx_operand): Make virtual and convert return
35125 type from void to rtx.
35126 (rtx_reader::read_until): New decl.
35127 (rtx_reader::handle_any_trailing_information): New virtual function.
35128 (rtx_reader::postprocess): New virtual function.
35129 (rtx_reader::finalize_string): New virtual function.
35130 (rtx_reader::m_in_call_function_usage): New field.
35131 (rtx_reader::m_reuse_rtx_by_id): New field.
35132 * read-rtl-function.c: New file.
35133 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35134 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35135 (selftest::verify_three_block_rtl_cfg): New decl.
35136 * read-rtl-function.h: New file.
35137 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35138 For host, include function.h, memmodel.h, and emit-rtl.h.
35139 (one_time_initialization): New function.
35140 (struct compact_insn_name): New struct.
35141 (compact_insn_names): New array.
35142 (find_code): Handle insn codes in compact dumps.
35143 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35144 (bind_subst_iter_and_attr): Likewise.
35145 (add_condition_to_string): Likewise.
35146 (add_condition_to_rtx): Likewise.
35147 (apply_attribute_uses): Likewise.
35148 (add_current_iterators): Likewise.
35149 (apply_iterators): Likewise.
35150 (initialize_iterators): Guard usage of apply_subst_iterator with
35151 #ifdef GENERATOR_FILE.
35152 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35153 (md_reader::read_mapping): Likewise.
35154 (add_define_attr_for_define_subst): Likewise.
35155 (add_define_subst_attr): Likewise.
35156 (read_subst_mapping): Likewise.
35157 (check_code_iterator): Likewise.
35158 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35160 (one_time_initialization): New function.
35161 (rtx_reader::read_until): New method.
35162 (read_flags): New function.
35163 (parse_reg_note_name): New function.
35164 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35165 Handle reuse_rtx ids.
35166 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35167 Add parsing support for RTL dumps, mirroring the special-cases in
35168 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35169 values, and calling handle_any_trailing_information.
35170 (rtx_reader::read_rtx_operand): Convert return type from void
35171 to rtx, returning return_rtx. Handle case 'e'. Call
35172 finalize_string on XSTR and XTMPL fields.
35173 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35174 "(nil)" values were omitted. Call the postprocess vfunc on the
35176 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35177 class ctor. Initialize m_in_call_function_usage. Call
35178 one_time_initialization.
35179 * rtl-tests.c (selftest::test_uncond_jump): Call
35180 set_new_first_and_last_insn.
35181 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35182 * selftest-rtl.c: New file.
35183 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35184 (selftest::get_insn_by_uid): New decl.
35185 * selftest-run-tests.c (selftest::run_tests): Call
35186 read_rtl_function_c_tests.
35187 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35188 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35191 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35193 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35194 operands in a special way. Assert that pos+len <= mode precision.
35196 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35198 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35199 3 argument Alias with unlimited for the negative form.
35200 (fno-vect-cost-model): Removed.
35202 2017-01-05 Martin Liska <mliska@suse.cz>
35204 * hsa-gen.c (gen_hsa_divmod): New function.
35205 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35207 2017-01-05 Martin Liska <mliska@suse.cz>
35210 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35213 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35215 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35216 small constant length operands.
35218 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35220 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35221 between loop iterations.
35223 2017-01-05 Martin Liska <mliska@suse.cz>
35226 * gimplify.c (gimplify_decl_expr): Compare to
35227 asan_poisoned_variables instread of checking flags.
35228 (gimplify_target_expr): Likewise.
35229 (gimplify_expr): Likewise.
35230 (gimplify_function_tree): Conditionally initialize
35231 asan_poisoned_variables.
35233 2017-01-04 Jeff Law <law@redhat.com>
35235 PR tree-optimizatin/78812
35236 * rtl.h (contains_mem_rtx_p): Prototype.
35237 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35238 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35239 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35240 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35241 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35243 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35245 * input.c (assert_char_at_range): Default-initialize actual_range.
35247 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35249 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35250 to match the caller.
35252 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35254 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35255 insns after final jump in test to emit dummy move.
35257 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35259 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35260 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35262 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35264 * multiple_target.c (create_dispatcher_calls): Init e_next.
35265 * tree-ssa-loop-split.c (split_loop): Init border.
35266 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35269 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35274 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35275 (altivec_register_operand): Do not return true if the operand
35276 contains a SUBREG mixing SImode and SFmode.
35277 (vsx_register_operand): Likewise.
35278 (vsx_reg_sfsubreg_ok): New predicate.
35279 (vfloat_operand): Do not return true if the operand contains a
35280 SUBREG mixing SImode and SFmode.
35281 (vint_operand): Likewise.
35282 (vlogical_operand): Likewise.
35283 (gpc_reg_operand): Likewise.
35284 (int_reg_operand): Likewise.
35285 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35286 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35287 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35289 (rs6000_emit_move_si_sf_subreg): New helper function.
35290 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35291 fixup SUBREGs involving SImode and SFmode.
35292 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35293 numbers for the new peephole2 optimization.
35294 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35295 the GLIBC math library that do AND/IOR/XOR operations on single
35296 precision floating point.
35297 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35298 target macros to say whether we need to avoid SUBREGs mixing
35300 (TARGET_ALLOW_SF_SUBREG): Likewise.
35301 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35302 (UNSPEC_SI_FROM_SF): Likewise.
35303 (iorxor): Change spacing.
35304 (and_ior_xor): New iterator for AND, IOR, and XOR.
35305 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35306 (movdi_from_sf_zero_ext): Likewise.
35307 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35308 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35309 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35310 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35311 (fms<mode>4): Likewise.
35312 (fnma<mode>4): Likewise.
35313 (fnms<mode>4): Likewise.
35314 (nfma<mode>4): Likewise.
35315 (nfms<mode>4): Likewise.
35317 2017-01-04 Marek Polacek <polacek@redhat.com>
35320 * doc/invoke.texi: Document -Wpointer-compare.
35322 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35324 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35327 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35328 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35331 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35334 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35335 documentation of the powerpc_popcntb_ok attribute.
35336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35337 code to issue warning messages if a requested CPU configuration is
35338 not supported by the binary (assembler and loader) toolchain.
35339 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35340 made to define a built-in function that has been disabled.
35341 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35342 made to define a built-in function that has been disabled.
35343 (altivec_init_builtins): Add comment explaining why definition
35344 of the DST built-in functions is not preceded by an assertion
35345 check. Add assertions to prevent ICE if attempts are made to
35346 define an altivec predicate or an abs* built-in function that has
35348 (htm_init_builtins): Add comment explaining why definition of the
35349 htm built-in functions is not preceded by an assertion check.
35351 2017-01-04 Jeff Law <law@redhat.com>
35353 PR tree-optimizatin/67955
35354 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35355 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35356 the points-to solution does not include pt_null. Use DECL_PT_UID
35359 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35361 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35362 Use gen_int_mode instead of gen_lopwart for const_int operands.
35364 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35366 PR tree-optimization/71563
35367 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35368 out of range value - has low bits known to be zero.
35370 2017-01-04 Alan Modra <amodra@gmail.com>
35372 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35373 * configure: Regenerate.
35374 * config.in: Regenerate.
35376 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35379 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35380 a substring of the message, but strcmp with the whole message. Ifdef
35381 ENABLE_NLS, translate the message first using dgettext.
35383 2017-01-03 Jeff Law <law@redhat.com>
35385 PR tree-optimizatin/78856
35386 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35387 (mark_threaded_blocks): Remove code to truncate thread paths that
35388 cross multiple loop headers. Instead invalidate the cached loop
35389 iteration information and handle case of a thread path walking
35390 into an irreducible region.
35392 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35395 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35396 assertions. Add support for doing the signbit if the IEEE 128-bit
35397 floating point value is in a GPR.
35398 * config/rs6000/rs6000.md (Fsignbit): Delete.
35399 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35400 Update the length attribute if the value is in a GPR.
35401 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35402 the sign or zero extension instruction, since the value is always 0/1.
35403 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35406 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35407 extracting SImode to a GPR register so that we can generate a
35408 store, limit the vector to be in a traditional Altivec register
35409 for the vextuwrx instruction.
35411 2017-01-03 Ian Lance Taylor <iant@google.com>
35413 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35415 2017-01-03 Martin Sebor <msebor@redhat.com>
35417 PR tree-optimization/78696
35418 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35419 precision. Use MPFR for %f for greater fidelity. Correct handling
35421 (pass_sprintf_length::compute_format_length): Set width and precision
35422 specified by asrerisk to void_node for vararg functions.
35423 (try_substitute_return_value): Adjust dump output.
35425 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35427 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35429 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35431 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35432 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35433 -mlra/-mno-lra was passed to the compiler.
35435 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35437 PR rtl-optimization/65618
35438 * emit-rtl.c (try_split): Move initialization of "before" and
35439 "after" to just before the call to emit_insn_after_setloc.
35441 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35443 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35445 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35447 * dwarf2out.c (gen_enumeration_type_die): When
35448 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35450 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35452 PR tree-optimization/78965
35453 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35454 Change first argument from const call_info & to call_info &. For %n
35455 set info.nowrite to false.
35457 PR middle-end/78901
35458 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35459 possibly throwing calls.
35461 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35462 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35463 and fns handling, rather than in a separate case SSA_NAME.
35465 2017-01-02 Jeff Law <law@redhat.com>
35467 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35468 fixes for first_period and second_period variables.
35470 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35473 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35474 (*insvqi_1): New insn pattern.
35475 (*insvqi_1_mem_rex64): Ditto.
35476 (*insvqi_2): Ditto.
35477 (*insvqi_3): Rename from *insvqi.
35479 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35481 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35483 * doc/cfg.texi (Edges): Remove reference to Java.
35484 (Maintaining the CFG): Ditto.
35486 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35488 PR middle-end/77674
35489 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35490 transparent aliases.
35492 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35494 PR middle-end/77484
35495 * predict.def (PRED_CALL): Update hitrate.
35496 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35497 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35498 into direct/indirect/polymorphic variants.
35500 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35502 Update copyright years.
35504 * gcc.c (process_command): Update copyright notice dates.
35505 * gcov-dump.c (print_version): Ditto.
35506 * gcov.c (print_version): Ditto.
35507 * gcov-tool.c (print_version): Ditto.
35508 * gengtype.c (create_file): Ditto.
35509 * doc/cpp.texi: Bump @copying's copyright year.
35510 * doc/cppinternals.texi: Ditto.
35511 * doc/gcc.texi: Ditto.
35512 * doc/gccint.texi: Ditto.
35513 * doc/gcov.texi: Ditto.
35514 * doc/install.texi: Ditto.
35515 * doc/invoke.texi: Ditto.
35517 Copyright (C) 2017 Free Software Foundation, Inc.
35519 Copying and distribution of this file, with or without modification,
35520 are permitted in any medium without royalty provided the copyright
35521 notice and this notice are preserved.